LIME [Local Interpretable Model agnostic Explainations]
\n\n##### LIME [Local Interpretable Model agnostic Explainations] where Local means local neighbourbhood of the instance Interpretable means a human should be able to interpret, model agnostic means which are applicable for all models and explaination means explaination that helps the interpretation.\n\n##### Before moving further lets see what is global and local interpretability?\n\n#### Kaggle allows less than 1MB ipynb file so I had to upload all output file along with better explaination over colab so kindly checkout the link below\nhttps://colab.research.google.com/drive/1PlngynWWBAbIrof9VjvkD8WJ59I3pdYG?usp=sharing\n\n\n##### What is the feature that differentiates these two guys A and B? why does the global model classify the point below B as unhealthy?\nFor A exercise duration makes the difference with its below point and in B body weight makes the main difference.\n\n###### Also for text data we can see\n\n| Text | Class |\n|----------------------------------------------------------------------------|---------------|\n| I have a `medical` `emergency`. Hence `Wont` be able to `attend` the `meeting` today | not important |\n| hi may i get `information` about your `service`? | important |\n\n##### Similarly for image data too as it mentions the part of image which made it classify to a particular category.\n\n\n\n###### Let's explore the intuition behind LIME. Linear models are easy to interpret because they are inherently linear. In such models, the equation is represented as:\n\n$y = w_1x_1 + w_2x_2 + w_3x_3$\n\n\nHere, $( w1x1 )$ represents the contribution of feature $( x1 )$ to the prediction for a specific data instance. A higher value of $( w1x1 )$ indicates a greater impact of that feature on the prediction.\n\n#### LIME works in 4 steps, which help explain predictions of complex, non-linear models locally.\n\n\n\n### Step 1: Local and Global Models\nThe first step shows a complex non-linear model where the classification of data points (e.g., \"Diabetes\" and \"No Diabetes\") is plotted based on features like body weight and exercise duration. The diagram highlights that it’s difficult to interpret why a specific point is classified in a certain way due to the model's complexity.\n\n### Step 2: Local Linear Approximation\n- In this step, LIME simplifies the complex non-linear model by locally approximating the area around the data point of interest. Instead of trying to interpret the entire model, LIME creates a linear model only for the points close to the selected instance. This makes the model easier to understand locally.\n\n### Step 3: Perturbed Data and Distance Weighting\n- This step shows how LIME generates perturbed (slightly altered) data points around the selected instance and weighs them according to their proximity. Random data points around the selected instance are generated, and their distances from the original point are measured. The idea is that features near the selected instance have more weight and influence, helping interpret which feature is more sensitive.\n\nFor example, in the diagram, the X-axis (body weight) is marked as sensitive since slight changes to this feature can flip the classification outcome.\n\n### Final Step: Feature Importance\n- In the last step, LIME generalizes the importance of each feature by applying the same local approximation strategy across multiple features. It calculates how important different features (e.g., age, blood pressure, weight, gender) are to the specific prediction. This provides an interpretable explanation of the model’s behavior for the selected data point.\n\n### Summary\nLIME works by:\n1. Distinguishing between local and global models to avoid complex global interpretation.\n2. Creating a simple, interpretable linear model near the selected data point.\n3. Using perturbed data points and distance-based weighting to understand feature sensitivity.\n4. Repeating the process to assess the importance of all features relevant to the specific instance.\n\n\nIn the image, we generate perturbed instances by highlighting superpixels with the highest positive weights as the explanation, while graying out all other areas.\n\n### NOTE: kaggle allows less than 1MB ipynb file so I had to upload all output file along with better explaination over colab so kindly checkout the link below\nhttps://colab.research.google.com/drive/1PlngynWWBAbIrof9VjvkD8WJ59I3pdYG?usp=sharing","metadata":{},"attachments":{"1c216fb2-4d92-4d56-bdc0-221d714b5cff.png":{"image/png":"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"}}},{"cell_type":"markdown","source":"\n#Mathematical Representation of LIME
\n### Original Representation of Data\n\n$$\n\\mathbf{x} \\in \\mathbb{R}^d\n$$\n\n### Transformed Representation\n\n$$\n\\mathbf{x}' = h^{-1}(h(\\mathbf{x}))\n$$\n\n### More Interpretable Representation or rather human interpretable representation of the instance (segments)\n\n$$\n\\mathbf{x}' \\in \\{0, 1\\}^d\n$$\n\n### Optimization Problem\n\n$$\n\\Xi(\\mathbf{x}) = \\arg\\min_{q \\in \\mathcal{L}} \\left[ \\sum_{i} L(f_i, g_i, \\pi_i, \\mathbf{x}) + \\lambda(q) \\right]\n$$\n\nWhere:\n- $( \\Xi(\\mathbf{x}) )$ is the optimization function.\n- $( q \\in \\mathcal{L} )$ represents a family of interpretable models.\n- $( \\sum_{i} L(f_i, g_i, \\pi_i, \\mathbf{x}) )$ represents a sum over the prediction errors (loss function).\n- $( \\lambda(q) )$ is a complexity measure for model \\( q \\).\n\n### Complexity Measure Equation\n\n$$\nu = u' \\left( \\frac{d}{d'} \\right)\n$$\n\nWhere,\n\n- $ \\mathbf{x} \\in \\mathbb{R}^d $ : Original representation of the data in d-dimensional space.\n- $ h(\\cdot) $ : Transformation function (e.g., logarithmic or other transformations).\n- $ \\mathbf{x}' \\in \\{0, 1\\}^d $ : Transformed representation, possibly interpretable for human analysis.\n- $ \\Xi(\\mathbf{x}) $ : The optimized model.\n- $ q \\in \\mathcal{L} $ : Family of interpretable models.\n- $ \\sum_{i} L(f_i, g_i, \\pi_i, \\mathbf{x}) $ : Sum of loss functions over all observations.\n- $ \\lambda(q) $ : Complexity measure of model \\( q \\).\n- $ \\mathbf{x} $ : The predicted instance we want to explain.\n- $ f $ : The complex model being explained.\n- $ g $ : The simple surrogate model.\n- $ \\pi_i $ : The neighborhood of $ \\mathbf{x} $.\n- $ \\lambda(q) $ : The complexity measure of $ g $.\n\n## Now this becomes a whole new problem let us see how we optimize it\n\n\n### 1. Minimizing the Loss Function:\n- The goal of LIME is to find a **simple surrogate model** $( g )$ that approximates the predictions of a **complex model** $( f )$ in a local neighborhood around an instance $( \\mathbf{x} )$.\n\n- The **loss function** that LIME tries to minimize is written as:\n $$\n L(f, g, \\pi_n) \\quad \\text{Local function L}\n $$\n This represents the **local loss** between the predictions of the complex model \\( f \\) and the surrogate model \\( g \\), weighted by the local neighborhood defined by \\( \\pi_n \\). \n\n### 2. Local Behavior of the Complex Model:\n- LIME focuses on approximating the behavior of the complex model \\( f \\) in a **local neighborhood** of - $( f(z) )$ is the prediction of the **complex model** for the perturbed instance $( z )$.\n - $( f(z) )$ is the prediction of the **complex model** for the perturbed instance ( z ).\n - $( g(z) )$ is the prediction of the **simple surrogate model** for the same instance ( z ).\n - $( \\pi_n(z) )$ is a weight assigned to each perturbation ( z ), representing its importance in the neighborhood of $( \\mathbf{x} )$.\n - The goal is to **minimize** the difference between the complex model’s predictions $(f(z))$ and the surrogate model’s predictions \\( g(z) \\), giving more importance to points that are closer to $( \\mathbf{x} )$ (i.e., ( z ) is close to $(\\mathbf{x}))$.\n\n- The weighted loss is given by:\n $$\n L(f, g, \\pi_n) = \\sum_{z \\in Z} \\pi_n(z) \\left( f(z) - g(z) \\right)^2\n $$\n\n\n### 3. Weighting Function for Local Samples:\n- The weight assigned to each perturbation \\( z \\) is determined by its distance from the instance \\( \\mathbf{x} \\), using a kernel function:\n $$\n \\pi_n(z) = \\exp\\left( \\frac{-D(\\mathbf{x}, z)^2}{\\sigma} \\right)\n $$\n\n Where:\n - $( D(\\mathbf{x}, z) )$ is the **Euclidean distance** between the instance $( \\mathbf{x} )$ and the perturbed sample $( z )$.\n - $( \\sigma )$ is the **kernel width** (a parameter that controls how fast the weight decays with distance).\n\n- This formula ensures that samples $( z )$ that are **closer to $( \\mathbf{x} )$** get a higher weight (i.e., they are more important for fitting the surrogate model), while samples that are further away receive smaller weights.\n\n","metadata":{}},{"cell_type":"code","source":"!pip install lime\n!pip install transformers shap datasets","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:05:31.710736Z","iopub.execute_input":"2024-10-03T10:05:31.711376Z","iopub.status.idle":"2024-10-03T10:06:00.949461Z","shell.execute_reply.started":"2024-10-03T10:05:31.711321Z","shell.execute_reply":"2024-10-03T10:06:00.948060Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"\n#LIME on TEXT DATA
","metadata":{}},{"cell_type":"code","source":"# Import necessary libraries\n\nimport pandas as pd\nfrom sklearn.model_selection import train_test_split\nimport matplotlib.pyplot as plt\nimport seaborn as sns\nfrom sklearn.pipeline import make_pipeline\nfrom lime.lime_text import LimeTextExplainer\nfrom sklearn.linear_model import LogisticRegression\nfrom sklearn.feature_extraction.text import TfidfVectorizer\nfrom collections import OrderedDict","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:06:00.951942Z","iopub.execute_input":"2024-10-03T10:06:00.952319Z","iopub.status.idle":"2024-10-03T10:06:02.664308Z","shell.execute_reply.started":"2024-10-03T10:06:00.952272Z","shell.execute_reply":"2024-10-03T10:06:02.663123Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Load training data\ntrain_df = pd.read_csv(\"/kaggle/input/quora-insincere-questions-classification/train.csv\")\nprint(\"Train shape : \", train_df.shape)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:06:02.665959Z","iopub.execute_input":"2024-10-03T10:06:02.666621Z","iopub.status.idle":"2024-10-03T10:06:07.727152Z","shell.execute_reply.started":"2024-10-03T10:06:02.666561Z","shell.execute_reply":"2024-10-03T10:06:07.726048Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_df.head()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:06:07.730445Z","iopub.execute_input":"2024-10-03T10:06:07.730947Z","iopub.status.idle":"2024-10-03T10:06:07.755784Z","shell.execute_reply.started":"2024-10-03T10:06:07.730891Z","shell.execute_reply":"2024-10-03T10:06:07.754353Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Display rows with NaN values\nnan_rows = train_df[train_df.isna().any(axis=1)]\nprint(nan_rows)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:06:07.757109Z","iopub.execute_input":"2024-10-03T10:06:07.757469Z","iopub.status.idle":"2024-10-03T10:06:08.053707Z","shell.execute_reply.started":"2024-10-03T10:06:07.757432Z","shell.execute_reply":"2024-10-03T10:06:08.052388Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Remove rows with NaN values\ntrain_df = train_df.dropna()\nprint(\"Train shape : \", train_df.shape)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:06:08.055504Z","iopub.execute_input":"2024-10-03T10:06:08.055983Z","iopub.status.idle":"2024-10-03T10:06:08.395118Z","shell.execute_reply.started":"2024-10-03T10:06:08.055930Z","shell.execute_reply":"2024-10-03T10:06:08.394017Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Split the data into training and validation sets\ntrain_df, val_df = train_test_split(train_df, test_size=0.1, random_state=2018)\nval_df.head()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:06:08.396458Z","iopub.execute_input":"2024-10-03T10:06:08.396794Z","iopub.status.idle":"2024-10-03T10:06:08.810852Z","shell.execute_reply.started":"2024-10-03T10:06:08.396758Z","shell.execute_reply":"2024-10-03T10:06:08.809760Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Select specific rows from validation set based on qid for inspection\ndf_select = pd.concat([val_df[val_df['qid'] == '1ffecf3a38aa5062f51c'], val_df[val_df['qid'] == 'd61b098340966d9d6501']], axis=0)\ndf_select.question_text","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:06:08.812670Z","iopub.execute_input":"2024-10-03T10:06:08.813153Z","iopub.status.idle":"2024-10-03T10:06:08.898715Z","shell.execute_reply.started":"2024-10-03T10:06:08.813093Z","shell.execute_reply":"2024-10-03T10:06:08.897345Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Reset index of validation dataframe\nval_df.reset_index(drop=True, inplace=True)\nval_df.head()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:06:08.900515Z","iopub.execute_input":"2024-10-03T10:06:08.900978Z","iopub.status.idle":"2024-10-03T10:06:08.913904Z","shell.execute_reply.started":"2024-10-03T10:06:08.900930Z","shell.execute_reply":"2024-10-03T10:06:08.912750Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Create a TF-IDF vectorizer and transform the training and validation data\n\n## vectorize to tf-idf vectors\ntfidf_vc = TfidfVectorizer(min_df = 10, max_features = 100000, analyzer = \"word\", ngram_range = (1, 2), stop_words = 'english', lowercase = True)\ntrain_vc = tfidf_vc.fit_transform(train_df[\"question_text\"])\nval_vc = tfidf_vc.transform(val_df[\"question_text\"])","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:06:08.919388Z","iopub.execute_input":"2024-10-03T10:06:08.919813Z","iopub.status.idle":"2024-10-03T10:07:10.162874Z","shell.execute_reply.started":"2024-10-03T10:06:08.919772Z","shell.execute_reply":"2024-10-03T10:07:10.161667Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Train a Logistic Regression model on the training data\n\nmodel = LogisticRegression(C = 0.5, solver = \"sag\")\nmodel = model.fit(train_vc, train_df.target)\n\n# Predict on the validation data\nval_pred = model.predict(val_vc)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:07:10.164283Z","iopub.execute_input":"2024-10-03T10:07:10.164650Z","iopub.status.idle":"2024-10-03T10:07:30.316248Z","shell.execute_reply.started":"2024-10-03T10:07:10.164610Z","shell.execute_reply":"2024-10-03T10:07:30.315019Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Calculate evaluation metrics\nfrom sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score, confusion_matrix, classification_report\n\naccuracy = accuracy_score(val_df.target, val_pred)\nprecision = precision_score(val_df.target, val_pred)\nrecall = recall_score(val_df.target, val_pred)\nf1 = f1_score(val_df.target, val_pred)\n\nprint(\"Accuracy:\", accuracy)\nprint(\"Precision:\", precision)\nprint(\"Recall:\", recall)\nprint(\"F1 Score:\", f1)\n\nconf_matrix = confusion_matrix(val_df.target, val_pred)\nprint(\"Confusion Matrix:\\n\", conf_matrix)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:07:30.317870Z","iopub.execute_input":"2024-10-03T10:07:30.318341Z","iopub.status.idle":"2024-10-03T10:07:30.497329Z","shell.execute_reply.started":"2024-10-03T10:07:30.318289Z","shell.execute_reply":"2024-10-03T10:07:30.496275Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Display classification report\n# Define class names\nclass_names = [\"sincere\", \"insincere\"]\nclass_report = classification_report(val_df.target, val_pred, target_names=class_names)\nprint(\"Classification Report:\\n\", class_report)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:07:30.498526Z","iopub.execute_input":"2024-10-03T10:07:30.498900Z","iopub.status.idle":"2024-10-03T10:07:30.704861Z","shell.execute_reply.started":"2024-10-03T10:07:30.498862Z","shell.execute_reply":"2024-10-03T10:07:30.703723Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Filter the rows where target is 1\ntarget_1_rows = val_df[val_df['target'] == 1]\n\n# Print the filtered rows and their row indices\nprint(\"Rows with target = 1:\")\ntarget_1_rows.head()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:07:30.706253Z","iopub.execute_input":"2024-10-03T10:07:30.706612Z","iopub.status.idle":"2024-10-03T10:07:30.723620Z","shell.execute_reply.started":"2024-10-03T10:07:30.706574Z","shell.execute_reply":"2024-10-03T10:07:30.722516Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"y=target_1_rows.index.tolist()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:07:30.725200Z","iopub.execute_input":"2024-10-03T10:07:30.725652Z","iopub.status.idle":"2024-10-03T10:07:30.731735Z","shell.execute_reply.started":"2024-10-03T10:07:30.725597Z","shell.execute_reply":"2024-10-03T10:07:30.730652Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Select a specific instance from the validation set for explanation\nimport numpy as np\nprediction_index = 20\nidx = int(val_df.index[prediction_index])\n# print(idx)\nc = make_pipeline(tfidf_vc, model)\nclass_names = [\"sincere\", \"insincere\"]\n\n# Create a LIME text explainer\nexplainer = LimeTextExplainer(class_names = class_names)\n\n# Explain the prediction for the selected instance\nexp = explainer.explain_instance(val_df[\"question_text\"][idx], c.predict_proba, num_features = 10)\n\n# Print the selected question text and its prediction probabilities\nprint(val_df[\"question_text\"][idx])\nprint(\"Probability (Insincere) =\", c.predict_proba([val_df[\"question_text\"][idx]])[0, 1])\nprint(\"Probability (Sincere) =\", c.predict_proba([val_df[\"question_text\"][idx]])[0, 0])\nprint(\"True Class is:\", class_names[int(val_df[\"target\"][idx])])","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:07:30.733756Z","iopub.execute_input":"2024-10-03T10:07:30.734216Z","iopub.status.idle":"2024-10-03T10:07:31.124366Z","shell.execute_reply.started":"2024-10-03T10:07:30.734166Z","shell.execute_reply":"2024-10-03T10:07:31.121501Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Get explanation weights as a list of tuples\nexp.as_list()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:07:31.133848Z","iopub.execute_input":"2024-10-03T10:07:31.137865Z","iopub.status.idle":"2024-10-03T10:07:31.157441Z","shell.execute_reply.started":"2024-10-03T10:07:31.137763Z","shell.execute_reply":"2024-10-03T10:07:31.155679Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Print original prediction probability\nprint('Original prediction:', model.predict_proba(val_vc[prediction_index])[0, 1])\n\n# Create a copy of the selected instance's TF-IDF vector and modify specific features\ntmp = val_vc[prediction_index].copy()\ntmp[0, tfidf_vc.vocabulary_['indians']] = 0\ntmp[0, tfidf_vc.vocabulary_['europeans']] = 0\n\n# Print prediction after removing specific features\nprint('Prediction after removing some features:', model.predict_proba(tmp)[0, 1])\n\n# Print the difference in prediction probabilities\nprint('Difference:', model.predict_proba(tmp)[0, 1] - model.predict_proba(val_vc[prediction_index])[0, 1])","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:07:31.158946Z","iopub.execute_input":"2024-10-03T10:07:31.159625Z","iopub.status.idle":"2024-10-03T10:07:31.177626Z","shell.execute_reply.started":"2024-10-03T10:07:31.159575Z","shell.execute_reply":"2024-10-03T10:07:31.175410Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Display LIME explanation in a notebook\nexp.show_in_notebook(text=val_df[\"question_text\"][idx], labels=(1,))","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:07:31.179339Z","iopub.execute_input":"2024-10-03T10:07:31.180217Z","iopub.status.idle":"2024-10-03T10:07:31.260422Z","shell.execute_reply.started":"2024-10-03T10:07:31.180165Z","shell.execute_reply":"2024-10-03T10:07:31.258920Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Extract and plot LIME weights\nweights = OrderedDict(exp.as_list())\nlime_weights = pd.DataFrame({\"words\": list(weights.keys()), \"weights\": list(weights.values())})\n\n# Plot the feature weights\nsns.barplot(x = \"words\", y = \"weights\", data = lime_weights, palette=\"viridis\")\nplt.xticks(rotation = 45)\nplt.title(\"Sample {} features weights given by LIME\".format(idx))\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:07:31.262125Z","iopub.execute_input":"2024-10-03T10:07:31.262597Z","iopub.status.idle":"2024-10-03T10:07:31.638549Z","shell.execute_reply.started":"2024-10-03T10:07:31.262546Z","shell.execute_reply":"2024-10-03T10:07:31.637578Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"\n\n#LIME on Image
","metadata":{}},{"cell_type":"code","source":"import os\nimport keras\nfrom keras.applications import inception_v3 as inc_net\nfrom keras.preprocessing import image\nfrom keras.applications.imagenet_utils import decode_predictions\nfrom skimage.io import imread\nimport matplotlib.pyplot as plt\n%matplotlib inline\nimport numpy as np\nprint('Notebook run using keras:', keras.__version__)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:07:31.639743Z","iopub.execute_input":"2024-10-03T10:07:31.640101Z","iopub.status.idle":"2024-10-03T10:07:46.529050Z","shell.execute_reply.started":"2024-10-03T10:07:31.640064Z","shell.execute_reply":"2024-10-03T10:07:46.527872Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"inet_model = inc_net.InceptionV3()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:07:46.530269Z","iopub.execute_input":"2024-10-03T10:07:46.530933Z","iopub.status.idle":"2024-10-03T10:07:49.477151Z","shell.execute_reply.started":"2024-10-03T10:07:46.530892Z","shell.execute_reply":"2024-10-03T10:07:49.476072Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def transform_img_fn(path_list):\n out = []\n for img_path in path_list:\n img = image.load_img(img_path, target_size=(299, 299))\n x = image.img_to_array(img)\n x = np.expand_dims(x, axis=0)\n x = inc_net.preprocess_input(x)\n out.append(x)\n return np.vstack(out)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:07:49.479570Z","iopub.execute_input":"2024-10-03T10:07:49.480039Z","iopub.status.idle":"2024-10-03T10:07:49.487015Z","shell.execute_reply.started":"2024-10-03T10:07:49.479988Z","shell.execute_reply":"2024-10-03T10:07:49.485862Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"images = transform_img_fn(['/kaggle/input/cat-jpeg/cat.jpeg']) \n#images = transform_img_fn([os.path.join('data','Tiger-1.jpg')])\n# I'm dividing by 2 and adding 0.5 because of how this Inception represents images\nplt.imshow(images[0] / 2 + 0.5)\npreds = inet_model.predict(images)\nfor x in decode_predictions(preds)[0]:\n print(x)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:07:49.488300Z","iopub.execute_input":"2024-10-03T10:07:49.488679Z","iopub.status.idle":"2024-10-03T10:07:53.253638Z","shell.execute_reply.started":"2024-10-03T10:07:49.488641Z","shell.execute_reply":"2024-10-03T10:07:53.252588Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"%load_ext autoreload\n%autoreload 2\nimport os,sys\ntry:\n import lime\nexcept:\n sys.path.append(os.path.join('..', '..')) # add the current directory\n import lime","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:07:53.254993Z","iopub.execute_input":"2024-10-03T10:07:53.255326Z","iopub.status.idle":"2024-10-03T10:07:53.341463Z","shell.execute_reply.started":"2024-10-03T10:07:53.255290Z","shell.execute_reply":"2024-10-03T10:07:53.340453Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from lime import lime_image\nexplainer = lime_image.LimeImageExplainer()\n\nexplanation = explainer.explain_instance(images[0].astype('double'), inet_model.predict, top_labels=5, hide_color=0, num_samples=1000)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:07:53.342850Z","iopub.execute_input":"2024-10-03T10:07:53.343254Z","iopub.status.idle":"2024-10-03T10:10:02.793364Z","shell.execute_reply.started":"2024-10-03T10:07:53.343214Z","shell.execute_reply":"2024-10-03T10:10:02.791777Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from skimage.segmentation import mark_boundaries\n\ntemp, mask = explanation.get_image_and_mask(explanation.top_labels[0], positive_only=True, num_features=5, hide_rest=True)\nplt.imshow(mark_boundaries(temp / 2 + 0.5, mask))","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:02.795446Z","iopub.execute_input":"2024-10-03T10:10:02.796690Z","iopub.status.idle":"2024-10-03T10:10:03.187619Z","shell.execute_reply.started":"2024-10-03T10:10:02.796635Z","shell.execute_reply":"2024-10-03T10:10:03.186369Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"temp, mask = explanation.get_image_and_mask(explanation.top_labels[0], positive_only=True, num_features=8, hide_rest=False)\nplt.imshow(mark_boundaries(temp / 2 + 0.5, mask))","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:03.195713Z","iopub.execute_input":"2024-10-03T10:10:03.196101Z","iopub.status.idle":"2024-10-03T10:10:03.657695Z","shell.execute_reply.started":"2024-10-03T10:10:03.196062Z","shell.execute_reply":"2024-10-03T10:10:03.656362Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"temp, mask = explanation.get_image_and_mask(explanation.top_labels[0], positive_only=False, num_features=10, hide_rest=False)\nplt.imshow(mark_boundaries(temp / 2 + 0.5, mask))","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:03.659229Z","iopub.execute_input":"2024-10-03T10:10:03.659684Z","iopub.status.idle":"2024-10-03T10:10:04.071491Z","shell.execute_reply.started":"2024-10-03T10:10:03.659630Z","shell.execute_reply":"2024-10-03T10:10:04.070426Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#Select the same class explained on the figures above.\nind = explanation.top_labels[0]\n\n#Map each explanation weight to the corresponding superpixel\ndict_heatmap = dict(explanation.local_exp[ind])\nheatmap = np.vectorize(dict_heatmap.get)(explanation.segments)\n\n#Plot. The visualization makes more sense if a symmetrical colorbar is used.\nplt.imshow(heatmap, cmap = 'RdBu', vmin = -heatmap.max(), vmax = heatmap.max())\nplt.colorbar()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:04.073136Z","iopub.execute_input":"2024-10-03T10:10:04.073620Z","iopub.status.idle":"2024-10-03T10:10:04.577570Z","shell.execute_reply.started":"2024-10-03T10:10:04.073566Z","shell.execute_reply":"2024-10-03T10:10:04.576296Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"temp, mask = explanation.get_image_and_mask(explanation.top_labels[1], positive_only=True, num_features=6, hide_rest=True)\nplt.imshow(mark_boundaries(temp / 2 + 0.5, mask))","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:04.579300Z","iopub.execute_input":"2024-10-03T10:10:04.579778Z","iopub.status.idle":"2024-10-03T10:10:04.971853Z","shell.execute_reply.started":"2024-10-03T10:10:04.579727Z","shell.execute_reply":"2024-10-03T10:10:04.970666Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"temp, mask = explanation.get_image_and_mask(explanation.top_labels[1], positive_only=False, num_features=5, hide_rest=False)\nplt.imshow(mark_boundaries(temp / 2 + 0.5, mask))","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:04.973384Z","iopub.execute_input":"2024-10-03T10:10:04.973882Z","iopub.status.idle":"2024-10-03T10:10:05.434252Z","shell.execute_reply.started":"2024-10-03T10:10:04.973797Z","shell.execute_reply":"2024-10-03T10:10:05.433076Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"\n\n#LIME on Tabular data
","metadata":{}},{"cell_type":"code","source":"import sklearn\nimport sklearn.datasets\nimport sklearn.ensemble\nimport numpy as np\nimport lime\nimport lime.lime_tabular\nfrom __future__ import print_function\n\n# Set seed for reproducibility\nnp.random.seed(1)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:05.435770Z","iopub.execute_input":"2024-10-03T10:10:05.436270Z","iopub.status.idle":"2024-10-03T10:10:05.773070Z","shell.execute_reply.started":"2024-10-03T10:10:05.436213Z","shell.execute_reply":"2024-10-03T10:10:05.771891Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import pandas as pd\nfrom sklearn.datasets import load_iris\n\n# Load the Iris dataset from sklearn\niris_data = load_iris()\n\n# Create a DataFrame from the dataset\niris_df = pd.DataFrame(data=iris_data.data, columns=iris_data.feature_names)\n\n# Add the target column to the DataFrame\niris_df['target'] = iris_data.target\n\n# Display the head of the DataFrame\niris_df.head()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:05.774763Z","iopub.execute_input":"2024-10-03T10:10:05.775187Z","iopub.status.idle":"2024-10-03T10:10:05.870057Z","shell.execute_reply.started":"2024-10-03T10:10:05.775135Z","shell.execute_reply":"2024-10-03T10:10:05.868990Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"iris = sklearn.datasets.load_iris()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:05.871393Z","iopub.execute_input":"2024-10-03T10:10:05.871731Z","iopub.status.idle":"2024-10-03T10:10:05.946500Z","shell.execute_reply.started":"2024-10-03T10:10:05.871696Z","shell.execute_reply":"2024-10-03T10:10:05.945328Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train, test, labels_train, labels_test = sklearn.model_selection.train_test_split(iris.data, iris.target, train_size=0.80)\n\nrf = sklearn.ensemble.RandomForestClassifier(n_estimators=500)\nrf.fit(train, labels_train)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:05.947799Z","iopub.execute_input":"2024-10-03T10:10:05.948200Z","iopub.status.idle":"2024-10-03T10:10:07.100992Z","shell.execute_reply.started":"2024-10-03T10:10:05.948161Z","shell.execute_reply":"2024-10-03T10:10:07.099894Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score, confusion_matrix\n\n# Predict labels on the test set\npredicted_labels = rf.predict(test)\n\n# Calculate accuracy\naccuracy = accuracy_score(labels_test, predicted_labels)\n\n# # Calculate precision\nprecision = precision_score(labels_test, predicted_labels, average='weighted')\n\n# # Calculate recall\nrecall = recall_score(labels_test, predicted_labels, average='weighted')\n\n# # Calculate F1 score\nf1 = f1_score(labels_test, predicted_labels, average='weighted')\n\n# Display summary\nprint(\"Summary:\")\nprint(\"Accuracy: {:.2f}\".format(accuracy))\nprint(\"Precision: {:.2f}\".format(precision))\nprint(\"Recall: {:.2f}\".format(recall))\nprint(\"F1 Score: {:.2f}\".format(f1))","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:07.102463Z","iopub.execute_input":"2024-10-03T10:10:07.102817Z","iopub.status.idle":"2024-10-03T10:10:07.219387Z","shell.execute_reply.started":"2024-10-03T10:10:07.102780Z","shell.execute_reply":"2024-10-03T10:10:07.218168Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Display confusion matrix\nconf_matrix = confusion_matrix(labels_test, predicted_labels)\nprint(\"\\nConfusion Matrix:\")\nprint(conf_matrix)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:07.220936Z","iopub.execute_input":"2024-10-03T10:10:07.221444Z","iopub.status.idle":"2024-10-03T10:10:07.298883Z","shell.execute_reply.started":"2024-10-03T10:10:07.221392Z","shell.execute_reply":"2024-10-03T10:10:07.297600Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"explainer = lime.lime_tabular.LimeTabularExplainer(train, feature_names=iris.feature_names, class_names=iris.target_names, discretize_continuous=True)\n\nprint(iris.feature_names)\nprint(iris.target_names)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:07.300101Z","iopub.execute_input":"2024-10-03T10:10:07.300447Z","iopub.status.idle":"2024-10-03T10:10:07.380080Z","shell.execute_reply.started":"2024-10-03T10:10:07.300409Z","shell.execute_reply":"2024-10-03T10:10:07.378970Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# i = np.random.randint(0, test.shape[0])\ni=12\n\n#Explain the instance using LIME\nexp = explainer.explain_instance(test[i], rf.predict_proba, num_features=6, top_labels=1)\n\n#Visualize the explanation\nprint(test[i], predicted_labels[i])\nprint(iris.feature_names)\nexp.show_in_notebook(show_table=True, show_all=False)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:07.381542Z","iopub.execute_input":"2024-10-03T10:10:07.382411Z","iopub.status.idle":"2024-10-03T10:10:07.847406Z","shell.execute_reply.started":"2024-10-03T10:10:07.382362Z","shell.execute_reply":"2024-10-03T10:10:07.845913Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(test[12])","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:07.849377Z","iopub.execute_input":"2024-10-03T10:10:07.850660Z","iopub.status.idle":"2024-10-03T10:10:07.948602Z","shell.execute_reply.started":"2024-10-03T10:10:07.850587Z","shell.execute_reply":"2024-10-03T10:10:07.947339Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"i = np.random.randint(0, test.shape[0])\nprint(test.shape[0])","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:07.950444Z","iopub.execute_input":"2024-10-03T10:10:07.950919Z","iopub.status.idle":"2024-10-03T10:10:08.027101Z","shell.execute_reply.started":"2024-10-03T10:10:07.950867Z","shell.execute_reply":"2024-10-03T10:10:08.025969Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print( rf.predict_proba(test[i].reshape(1,-1)))","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:08.028711Z","iopub.execute_input":"2024-10-03T10:10:08.029566Z","iopub.status.idle":"2024-10-03T10:10:08.142495Z","shell.execute_reply.started":"2024-10-03T10:10:08.029511Z","shell.execute_reply":"2024-10-03T10:10:08.141382Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"feature_index = lambda x: iris.feature_names.index(x)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:08.144100Z","iopub.execute_input":"2024-10-03T10:10:08.145083Z","iopub.status.idle":"2024-10-03T10:10:08.219221Z","shell.execute_reply.started":"2024-10-03T10:10:08.145028Z","shell.execute_reply":"2024-10-03T10:10:08.217913Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"i= 12\n\ntemp = test[i].copy()\ntemp_before = test[i].copy()\n\ntemp[feature_index('petal width (cm)')] = temp[feature_index('petal width (cm)')] - 1.5\n\nprint('Petal width before:', temp_before[feature_index('petal width (cm)')])\nprint('Petal width after:', temp[feature_index('petal width (cm)')])\nprint ()\n\nprint('P(setosa) before:', rf.predict_proba(temp_before.reshape(1,-1))[0,0])\nprint('P(setosa) after:', rf.predict_proba(temp.reshape(1,-1))[0,0])\nprint ()\n\nprint('P(versicolor) before:', rf.predict_proba(temp_before.reshape(1,-1))[0,1])\nprint('P(versicolor) after:', rf.predict_proba(temp.reshape(1,-1))[0,1])\nprint ()\n\nprint('P(virginica) before:', rf.predict_proba(temp_before.reshape(1,-1))[0,2])\nprint('P(virginica) after:', rf.predict_proba(temp.reshape(1,-1))[0,2])\nprint ()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:08.221358Z","iopub.execute_input":"2024-10-03T10:10:08.221747Z","iopub.status.idle":"2024-10-03T10:10:08.497959Z","shell.execute_reply.started":"2024-10-03T10:10:08.221702Z","shell.execute_reply":"2024-10-03T10:10:08.496752Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import matplotlib.pyplot as plt\n\n# Make a copy of the instance\ntemp = test[i].copy()\ntemp_before = test[i].copy()\n\n# Increase petal length by 1\ntemp[feature_index('petal width (cm)')] -= 1.7\n\n# Predict probabilities before and after perturbation\nproba_before = rf.predict_proba(temp_before.reshape(1,-1))[0]\nproba_after = rf.predict_proba(temp.reshape(1,-1))[0]\n\n# Labels for the plot\nlabels = iris.target_names\n\n# Plotting\nfig, ax = plt.subplots()\nx = np.arange(len(labels))\nwidth = 0.35\n\n# Plot probabilities before and after perturbation\nrects1 = ax.bar(x - width/2, proba_before, width, label='Before')\nrects2 = ax.bar(x + width/2, proba_after, width, label='After')\n\n# Add labels, title, and legend\nax.set_xlabel('Classes')\nax.set_ylabel('Probability')\nax.set_title('Effect of changing Petal Width')\nax.set_xticks(x)\nax.set_xticklabels(labels)\nax.legend()\n\n# Display the plot\nprint('Petal length before:', temp_before[feature_index('petal length (cm)')])\nprint('Petal length after:', temp[feature_index('petal length (cm)')])\nprint ()\nprint('Petal width before:', temp_before[feature_index('petal width (cm)')])\nprint('Petal width after:', temp[feature_index('petal width (cm)')])\nprint ()\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:08.499626Z","iopub.execute_input":"2024-10-03T10:10:08.500097Z","iopub.status.idle":"2024-10-03T10:10:08.855456Z","shell.execute_reply.started":"2024-10-03T10:10:08.500047Z","shell.execute_reply":"2024-10-03T10:10:08.854320Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"temp = test[i].copy()\ntemp_before = test[i].copy()\n\ntemp[feature_index('petal length (cm)')] = temp[feature_index('petal length (cm)')] + 1\n\nprint('Petal length before:', temp_before[feature_index('petal length (cm)')])\nprint('Petal length after:', temp[feature_index('petal length (cm)')])\nprint ()\n\nprint('P(setosa) before:', rf.predict_proba(temp_before.reshape(1,-1))[0,0])\nprint('P(setosa) after:', rf.predict_proba(temp.reshape(1,-1))[0,0])\nprint ()\n\nprint('P(versicolor) before:', rf.predict_proba(temp_before.reshape(1,-1))[0,1])\nprint('P(versicolor) after:', rf.predict_proba(temp.reshape(1,-1))[0,1])\nprint ()\n\nprint('P(virginica) before:', rf.predict_proba(temp_before.reshape(1,-1))[0,2])\nprint('P(virginica) after:', rf.predict_proba(temp.reshape(1,-1))[0,2])\nprint ()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:08.857027Z","iopub.execute_input":"2024-10-03T10:10:08.857489Z","iopub.status.idle":"2024-10-03T10:10:09.128543Z","shell.execute_reply.started":"2024-10-03T10:10:08.857439Z","shell.execute_reply":"2024-10-03T10:10:09.127227Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import matplotlib.pyplot as plt\n\n# Make a copy of the instance\ntemp = test[i].copy()\ntemp_before = test[i].copy()\n\n# Increase petal length by 1\ntemp[feature_index('petal length (cm)')] -= 4.5\n\n# Predict probabilities before and after perturbation\nproba_before = rf.predict_proba(temp_before.reshape(1,-1))[0]\nproba_after = rf.predict_proba(temp.reshape(1,-1))[0]\n\n# Labels for the plot\nlabels = iris.target_names\n\n# Plotting\nfig, ax = plt.subplots()\nx = np.arange(len(labels))\nwidth = 0.35\n\n# Plot probabilities before and after perturbation\nrects1 = ax.bar(x - width/2, proba_before, width, label='Before')\nrects2 = ax.bar(x + width/2, proba_after, width, label='After')\n\n# Add labels, title, and legend\nax.set_xlabel('Classes')\nax.set_ylabel('Probability')\nax.set_title('Effect of changing Petal Length')\nax.set_xticks(x)\nax.set_xticklabels(labels)\nax.legend()\n\n# Display the plot\nprint('Petal length before:', temp_before[feature_index('petal length (cm)')])\nprint('Petal length after:', temp[feature_index('petal length (cm)')])\nprint ()\nprint('Petal width before:', temp_before[feature_index('petal width (cm)')])\nprint('Petal width after:', temp[feature_index('petal width (cm)')])\nprint ()\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:09.130157Z","iopub.execute_input":"2024-10-03T10:10:09.130614Z","iopub.status.idle":"2024-10-03T10:10:09.499989Z","shell.execute_reply.started":"2024-10-03T10:10:09.130562Z","shell.execute_reply":"2024-10-03T10:10:09.498906Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"temp = test[i].copy()\ntemp_before = test[i].copy()\n\ntemp[feature_index('petal length (cm)')] = temp[feature_index('petal length (cm)')] -1\ntemp[feature_index('petal width (cm)')] = temp[feature_index('petal width (cm)')] + 2\n\nprint('Petal length before:', temp_before[feature_index('petal length (cm)')])\nprint('Petal length after:', temp[feature_index('petal length (cm)')])\nprint ()\n\nprint('Petal width before:', temp_before[feature_index('petal width (cm)')])\nprint('Petal width after:', temp[feature_index('petal width (cm)')])\nprint ()\n\nprint('P(setosa) before:', rf.predict_proba(temp_before.reshape(1,-1))[0,0])\nprint('P(setosa) after:', rf.predict_proba(temp.reshape(1,-1))[0,0])\nprint ()\n\nprint('P(versicolor) before:', rf.predict_proba(temp_before.reshape(1,-1))[0,1])\nprint('P(versicolor) after:', rf.predict_proba(temp.reshape(1,-1))[0,1])\nprint ()\n\nprint('P(virginica) before:', rf.predict_proba(temp_before.reshape(1,-1))[0,2])\nprint('P(virginica) after:', rf.predict_proba(temp.reshape(1,-1))[0,2])\nprint ()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:09.501313Z","iopub.execute_input":"2024-10-03T10:10:09.501654Z","iopub.status.idle":"2024-10-03T10:10:09.764738Z","shell.execute_reply.started":"2024-10-03T10:10:09.501618Z","shell.execute_reply":"2024-10-03T10:10:09.763668Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import matplotlib.pyplot as plt\n\n# Make a copy of the instance\ntemp = test[i].copy()\ntemp_before = test[i].copy()\n\n# Increase petal length by 1\ntemp[feature_index('petal length (cm)')] -= 4.5\ntemp[feature_index('petal width (cm)')] -= 1.5\n\n# Predict probabilities before and after perturbation\nproba_before = rf.predict_proba(temp_before.reshape(1,-1))[0]\nproba_after = rf.predict_proba(temp.reshape(1,-1))[0]\n\n# Labels for the plot\nlabels = iris.target_names\n\n# Plotting\nfig, ax = plt.subplots()\nx = np.arange(len(labels))\nwidth = 0.35\n\n# Plot probabilities before and after perturbation\nrects1 = ax.bar(x - width/2, proba_before, width, label='Before')\nrects2 = ax.bar(x + width/2, proba_after, width, label='After')\n\n# Add labels, title, and legend\nax.set_xlabel('Classes')\nax.set_ylabel('Probability')\nax.set_title('Effect of changing Petal Length and Width')\nax.set_xticks(x)\nax.set_xticklabels(labels)\nax.legend()\n\n# Display the plot\nprint('Petal length before:', temp_before[feature_index('petal length (cm)')])\nprint('Petal length after:', temp[feature_index('petal length (cm)')])\nprint ()\nprint('Petal width before:', temp_before[feature_index('petal width (cm)')])\nprint('Petal width after:', temp[feature_index('petal width (cm)')])\nprint ()\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:09.766167Z","iopub.execute_input":"2024-10-03T10:10:09.766534Z","iopub.status.idle":"2024-10-03T10:10:10.121737Z","shell.execute_reply.started":"2024-10-03T10:10:09.766496Z","shell.execute_reply":"2024-10-03T10:10:10.120680Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"exp.show_in_notebook(show_table=True, show_all=True)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:10.123076Z","iopub.execute_input":"2024-10-03T10:10:10.123415Z","iopub.status.idle":"2024-10-03T10:10:10.229073Z","shell.execute_reply.started":"2024-10-03T10:10:10.123379Z","shell.execute_reply":"2024-10-03T10:10:10.227879Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"\n\n\n#Limitations of LIME
\n\n- **Local vs Global**:\n - **LIME** explains the behavior of the model **locally** for individual predictions by focusing on the neighborhood of $( \\mathbf{x} )$. While this provides useful insights for a specific instance, it **may not capture the global behavior** of the model. \n - In other words, LIME is designed to explain one prediction at a time, which might not generalize to other predictions or the overall decision boundaries of the model.\n\n- **Linear Assumption**:\n - LIME approximates the complex model using a **linear surrogate model**. This is a simplification, which works well for **small local neighborhoods**, but can lead to inaccuracies in more complex regions where the model behaves in a non-linear manner.\n - The **linear approximation** might not represent the true complexity of the model’s decision boundary, especially when the complex model uses highly non-linear relationships to make its predictions.","metadata":{}},{"cell_type":"markdown","source":"\n#SHAP [Shapley Additive Explainations]
\n\nSHAP (SHapley Additive exPlanations) is another Explainable AI method used to explain the output of machine learning models. It helps to interpret the predictions made by complex models by assigning a value to each feature based on its contribution to the prediction\n\nSHAP is based on a straightforward idea. Imagine a team of five people participating in a hackathon:\n\n1. The first person is skilled in machine learning (ML).\n2. The second person is an expert in backend development.\n3. The third person specializes in DevOps tools.\n4. The fourth person has excellent ML and communication skills.\n5. The fifth person is new to the team and helps with documentation.\n\nThe prize money is distributed as follows:\n- First place: 50,000 USD,\n- Second place: 30,000 USD,\n- Third place: 10,000 USD\n\nTo fairly distribute the prize money, we need to consider how each team member contributed since some may have worked harder than others. Instead of splitting the prize equally, we look at different combinations of team members to see their contributions.\n\nFor example:\n- If only the 1st, 2nd, 4th, and 5th members participated, they would earn the 3rd place prize.\n- If all team members participate, they would earn the 1st place prize.\n- If the 1st, 2nd, 3rd, and 5th members are involved, they would get the 2nd place prize.\n- If the 1st, 3rd, 4th, and 5th members are included, they wouldn't place at all.\n- If the 1st, 2nd, 3rd, and 4th members work together, they would again win 1st place.\n\nBy considering all possible combinations of team members, we can determine a fair way to distribute the prize money based on each member’s contribution.\n\nSo in short\n- To calculate the true individual contribution we need to consider different subsets.\n- We calculate the individual's contribution for each subset and then do an average of these contribution to find the marginal contribution of each player.\n\nImagine we are building a model to classify individuals as diabetic or non-diabetic. We have several features, including age, exercise duration, BMI, the presence of diabetic parents, and calorie intake. In this scenario, it's essential to determine which feature has the most significant impact on the model's predictions. Since we are dealing with specific cases, we need local interpretation rather than generalization.\n\nSo while creating subset we need to keep this thing in mind that the dimenions should be correct we we fill with random values in the missing feature\n\n| Features | 1 | 2 | 3 | 4 | 5 |\n|-----------------------|------|------|------|------|------|\n| Subset | 1 | | 3 | | 5 |\n| correct_dimensions_subset | 1 | Rand | 3 | Rand | 5 |\n\nMathematical intuition:\n\n$$\nQ_i(f_j = x) = \\sum \\frac{|Z'|! (M - |Z'| - 1)!}{M!} \\left[ f_u(Z') - f_x(Z' \\setminus \\{i\\}) \\right]\n$$\n\nwhere,\n\n- $( Q_i )$ be the **Shapley value** for feature $( i )$,\n- $( f )$ be the **ML model**,\n- $( x )$ be a **data point**,\n- $( Z' )$ be the **subsets of features**,\n- $( x' )$ is **simplified to** $( x' )$ **by mapping**,\n- $( \\left( \\sum \\frac{|Z'|! (M - |Z'| - 1)!}{M!} \\right) )$ is the **weighting term**, where $(M)$ is the **total number of features**, and $( Z' )$ is the **number of subsets with feature** $( i )$ and $(|Z'|!)$ is the factorial of the size of the subset $( Z' )$,$( (M - |Z'| - 1)! )$ is the factorial of the number of features not included in the subset and $( M!)$ is the factorial of the total number of features.\n- $( f_x(Z') )$ represents the prediction **with feature** $( i )$ of interest,\n- $( f_x(Z' \\setminus \\{i\\}) )$ represents the prediction **without feature** $( i )$ (the feature of interest).\n\nSome of the important plots\n- beesworm summary plot\n- dependence plot\n- heatmap\n- decision plot\n- imageplot\n- textplot\n- scatterplot\n- forceplot\n- waterfall plot\n- barplot\n\n","metadata":{}},{"cell_type":"markdown","source":"\n\n#SHAP for Text data
","metadata":{}},{"cell_type":"code","source":"import datasets\nimport numpy as np\nimport transformers\nimport shap","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:10.230600Z","iopub.execute_input":"2024-10-03T10:10:10.230999Z","iopub.status.idle":"2024-10-03T10:10:16.313645Z","shell.execute_reply.started":"2024-10-03T10:10:10.230960Z","shell.execute_reply":"2024-10-03T10:10:16.312352Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"dataset = datasets.load_dataset(\"imdb\", split=\"test\")\n\n# shorten the strings to fit into the pipeline model\nshort_data = [v[:500] for v in dataset[\"text\"][:20]]","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:16.315937Z","iopub.execute_input":"2024-10-03T10:10:16.317409Z","iopub.status.idle":"2024-10-03T10:10:24.422630Z","shell.execute_reply.started":"2024-10-03T10:10:16.317334Z","shell.execute_reply":"2024-10-03T10:10:24.421562Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(dataset[0])\n","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:24.424107Z","iopub.execute_input":"2024-10-03T10:10:24.424442Z","iopub.status.idle":"2024-10-03T10:10:24.515228Z","shell.execute_reply.started":"2024-10-03T10:10:24.424399Z","shell.execute_reply":"2024-10-03T10:10:24.513931Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(short_data[0])","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:24.516791Z","iopub.execute_input":"2024-10-03T10:10:24.517891Z","iopub.status.idle":"2024-10-03T10:10:24.609233Z","shell.execute_reply.started":"2024-10-03T10:10:24.517810Z","shell.execute_reply":"2024-10-03T10:10:24.608136Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import pandas as pd\n\n# Create a pandas dataframe\ndf = pd.DataFrame(short_data)\n\n# Display the head of the dataframe\ndf.head()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:24.610963Z","iopub.execute_input":"2024-10-03T10:10:24.611401Z","iopub.status.idle":"2024-10-03T10:10:24.709815Z","shell.execute_reply.started":"2024-10-03T10:10:24.611350Z","shell.execute_reply":"2024-10-03T10:10:24.708601Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"classifier = transformers.pipeline(\"sentiment-analysis\", return_all_scores=True)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:24.711356Z","iopub.execute_input":"2024-10-03T10:10:24.711817Z","iopub.status.idle":"2024-10-03T10:10:32.997042Z","shell.execute_reply.started":"2024-10-03T10:10:24.711766Z","shell.execute_reply":"2024-10-03T10:10:32.995791Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"texts = [\"I love this movie! It was the best movie of all time\", \"I liked the first half but did not like the second half.\"]\n\nresults = classifier(texts)\nprint(results)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:32.998707Z","iopub.execute_input":"2024-10-03T10:10:32.999683Z","iopub.status.idle":"2024-10-03T10:10:33.305976Z","shell.execute_reply.started":"2024-10-03T10:10:32.999621Z","shell.execute_reply":"2024-10-03T10:10:33.304880Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(\"1: \", short_data[0],\"\\n\")\nprint(\"2: \", short_data[1],\"\\n\")\nprint(\"3: \",short_data[2],\"\\n\")\nprint(\"\\n\")\nprint(classifier(short_data[:3]))","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:33.307366Z","iopub.execute_input":"2024-10-03T10:10:33.307715Z","iopub.status.idle":"2024-10-03T10:10:33.787697Z","shell.execute_reply.started":"2024-10-03T10:10:33.307677Z","shell.execute_reply":"2024-10-03T10:10:33.786603Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from sklearn.metrics import accuracy_score\n\n# Predict the sentiment for the short data\npredictions = classifier(short_data)\n\n# Convert the predictions to labels\npredicted_labels = []\nfor prediction in predictions:\n predicted_label = max(prediction, key=lambda x: x['score'])['label']\n predicted_labels.append(0 if predicted_label == 'NEGATIVE' else 1) # Assuming 'NEGATIVE' is 0 and 'POSITIVE' is 1\n\ntrue_labels = dataset[\"label\"][:20] # Get the true labels\n\n# Calculate the accuracy\naccuracy = accuracy_score(true_labels, predicted_labels)\nprint(f\"Accuracy: {accuracy}\")\n\nprint(true_labels)\n\nprint(predicted_labels)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:33.789282Z","iopub.execute_input":"2024-10-03T10:10:33.789846Z","iopub.status.idle":"2024-10-03T10:10:36.552872Z","shell.execute_reply.started":"2024-10-03T10:10:33.789775Z","shell.execute_reply":"2024-10-03T10:10:36.551649Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(dataset[12500])","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:36.554444Z","iopub.execute_input":"2024-10-03T10:10:36.554872Z","iopub.status.idle":"2024-10-03T10:10:36.672211Z","shell.execute_reply.started":"2024-10-03T10:10:36.554816Z","shell.execute_reply":"2024-10-03T10:10:36.671162Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# define the explainer\nexplainer = shap.Explainer(classifier)\nshap_values = explainer(short_data[:2])","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:10:36.673933Z","iopub.execute_input":"2024-10-03T10:10:36.674296Z","iopub.status.idle":"2024-10-03T10:12:57.221023Z","shell.execute_reply.started":"2024-10-03T10:10:36.674257Z","shell.execute_reply":"2024-10-03T10:12:57.219881Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(shap_values.shape)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:12:57.222512Z","iopub.execute_input":"2024-10-03T10:12:57.222928Z","iopub.status.idle":"2024-10-03T10:12:57.348555Z","shell.execute_reply.started":"2024-10-03T10:12:57.222886Z","shell.execute_reply":"2024-10-03T10:12:57.347270Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"shap.plots.text(shap_values[0]) # Index into a single instance\n","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:12:57.484735Z","iopub.execute_input":"2024-10-03T10:12:57.485234Z","iopub.status.idle":"2024-10-03T10:12:57.739875Z","shell.execute_reply.started":"2024-10-03T10:12:57.485174Z","shell.execute_reply":"2024-10-03T10:12:57.738632Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"shap.plots.text(shap_values[1]) # Index into a single instance","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:15:30.402423Z","iopub.execute_input":"2024-10-03T10:15:30.402888Z","iopub.status.idle":"2024-10-03T10:15:30.628388Z","shell.execute_reply.started":"2024-10-03T10:15:30.402843Z","shell.execute_reply":"2024-10-03T10:15:30.627062Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"masker = shap.maskers.Text(classifier.tokenizer)\nexplainer3 = shap.Explainer(pmodel, masker)\nshap_values3 = explainer3(short_data[:2])\nshap.plots.text(shap_values[0])","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:25:52.566080Z","iopub.execute_input":"2024-10-03T10:25:52.566558Z","iopub.status.idle":"2024-10-03T10:28:05.219552Z","shell.execute_reply.started":"2024-10-03T10:25:52.566515Z","shell.execute_reply":"2024-10-03T10:28:05.218344Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"masker.shape(\"I like this movie.\")\n\nmodel_args = masker(\n np.array([True, True, True, True, True, True, True]), \"I like this movie.\"\n)\nmodel_args","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:29:57.176251Z","iopub.execute_input":"2024-10-03T10:29:57.176694Z","iopub.status.idle":"2024-10-03T10:29:57.299109Z","shell.execute_reply.started":"2024-10-03T10:29:57.176656Z","shell.execute_reply":"2024-10-03T10:29:57.297914Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"pmodel(*model_args)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:30:00.720378Z","iopub.execute_input":"2024-10-03T10:30:00.720835Z","iopub.status.idle":"2024-10-03T10:30:00.897895Z","shell.execute_reply.started":"2024-10-03T10:30:00.720779Z","shell.execute_reply":"2024-10-03T10:30:00.896756Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"model_args = masker(\n np.array([True, True, False, False, True, True, True]), \"I like this movie.\"\n)\nmodel_args","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:30:02.924649Z","iopub.execute_input":"2024-10-03T10:30:02.925090Z","iopub.status.idle":"2024-10-03T10:30:03.045135Z","shell.execute_reply.started":"2024-10-03T10:30:02.925050Z","shell.execute_reply":"2024-10-03T10:30:03.043861Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"pmodel(*model_args)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:30:04.705277Z","iopub.execute_input":"2024-10-03T10:30:04.706149Z","iopub.status.idle":"2024-10-03T10:30:04.853301Z","shell.execute_reply.started":"2024-10-03T10:30:04.706101Z","shell.execute_reply":"2024-10-03T10:30:04.852209Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"masker2 = shap.maskers.Text(\n classifier.tokenizer, mask_token=\"...\", collapse_mask_token=True\n)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:30:06.794904Z","iopub.execute_input":"2024-10-03T10:30:06.795333Z","iopub.status.idle":"2024-10-03T10:30:06.916158Z","shell.execute_reply.started":"2024-10-03T10:30:06.795294Z","shell.execute_reply":"2024-10-03T10:30:06.915033Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"model_args2 = masker2(\n np.array([True, True, False, False, True, True, True]), \"I like this movie.\"\n)\nmodel_args2","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:30:09.114867Z","iopub.execute_input":"2024-10-03T10:30:09.115291Z","iopub.status.idle":"2024-10-03T10:30:09.237550Z","shell.execute_reply.started":"2024-10-03T10:30:09.115252Z","shell.execute_reply":"2024-10-03T10:30:09.236313Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"pmodel(*model_args2)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:30:11.390017Z","iopub.execute_input":"2024-10-03T10:30:11.390468Z","iopub.status.idle":"2024-10-03T10:30:11.541454Z","shell.execute_reply.started":"2024-10-03T10:30:11.390427Z","shell.execute_reply":"2024-10-03T10:30:11.540333Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# explain the predictions of the pipeline on the first two samples\nshap_values = explainer(short_data[:20])","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:30:13.305062Z","iopub.execute_input":"2024-10-03T10:30:13.305506Z","iopub.status.idle":"2024-10-03T10:52:34.722740Z","shell.execute_reply.started":"2024-10-03T10:30:13.305464Z","shell.execute_reply":"2024-10-03T10:52:34.721638Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"shap.plots.bar(shap_values[0, :, \"POSITIVE\"])","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:52:39.164788Z","iopub.execute_input":"2024-10-03T10:52:39.165414Z","iopub.status.idle":"2024-10-03T10:52:40.409541Z","shell.execute_reply.started":"2024-10-03T10:52:39.165347Z","shell.execute_reply":"2024-10-03T10:52:40.408487Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"\n\n\n#SHAP for Image data\n
","metadata":{}},{"cell_type":"code","source":"import json\nfrom tensorflow.keras.applications.resnet50 import ResNet50, preprocess_input\nimport shap\nimport numpy as np\nimport matplotlib.pyplot as plt\n\nmodel = ResNet50(weights=\"imagenet\")\nX, y = shap.datasets.imagenet50()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:54:07.990211Z","iopub.execute_input":"2024-10-03T10:54:07.990628Z","iopub.status.idle":"2024-10-03T10:54:11.712699Z","shell.execute_reply.started":"2024-10-03T10:54:07.990590Z","shell.execute_reply":"2024-10-03T10:54:11.711730Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(y)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:54:14.449770Z","iopub.execute_input":"2024-10-03T10:54:14.450554Z","iopub.status.idle":"2024-10-03T10:54:14.568042Z","shell.execute_reply.started":"2024-10-03T10:54:14.450510Z","shell.execute_reply":"2024-10-03T10:54:14.566914Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"plt.imshow(X[20])","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:54:17.210216Z","iopub.execute_input":"2024-10-03T10:54:17.210644Z","iopub.status.idle":"2024-10-03T10:54:17.640651Z","shell.execute_reply.started":"2024-10-03T10:54:17.210605Z","shell.execute_reply":"2024-10-03T10:54:17.639424Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Assuming X[8] contains integer data that needs to be scaled to the range [0, 255]\nX = np.clip(X, 0, 255).astype(np.uint8)\n\nplt.imshow(X[5])","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:54:25.534624Z","iopub.execute_input":"2024-10-03T10:54:25.535660Z","iopub.status.idle":"2024-10-03T10:54:26.003377Z","shell.execute_reply.started":"2024-10-03T10:54:25.535614Z","shell.execute_reply":"2024-10-03T10:54:26.002248Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(X.shape)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:54:30.369992Z","iopub.execute_input":"2024-10-03T10:54:30.371015Z","iopub.status.idle":"2024-10-03T10:54:30.488414Z","shell.execute_reply.started":"2024-10-03T10:54:30.370969Z","shell.execute_reply":"2024-10-03T10:54:30.487291Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# getting ImageNet 1000 class names\nurl = \"https://s3.amazonaws.com/deep-learning-models/image-models/imagenet_class_index.json\"\nwith open(shap.datasets.cache(url)) as file:\n class_names = [v[1] for v in json.load(file).values()]\n \nprint(\"Number of ImageNet classes:\", len(class_names))\nprint(\"Class names:\", class_names)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:54:34.639722Z","iopub.execute_input":"2024-10-03T10:54:34.640215Z","iopub.status.idle":"2024-10-03T10:54:34.918991Z","shell.execute_reply.started":"2024-10-03T10:54:34.640174Z","shell.execute_reply":"2024-10-03T10:54:34.917875Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(y[17])","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:54:40.230679Z","iopub.execute_input":"2024-10-03T10:54:40.231909Z","iopub.status.idle":"2024-10-03T10:54:40.350367Z","shell.execute_reply.started":"2024-10-03T10:54:40.231851Z","shell.execute_reply":"2024-10-03T10:54:40.349039Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def f(x):\n tmp = x.copy()\n preprocess_input(tmp)\n return model(tmp)\n\nmasker = shap.maskers.Image(\"inpaint_telea\", X[0].shape)\n\nexplainer = shap.Explainer(f, masker, output_names=class_names)\n\nshap_values = explainer(\n X[1:3], max_evals=100, batch_size=50, outputs=shap.Explanation.argsort.flip[:4]\n)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:54:42.569980Z","iopub.execute_input":"2024-10-03T10:54:42.570472Z","iopub.status.idle":"2024-10-03T10:55:34.611075Z","shell.execute_reply.started":"2024-10-03T10:54:42.570426Z","shell.execute_reply":"2024-10-03T10:55:34.603937Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# output with shap values\nshap.image_plot(shap_values)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:55:34.620182Z","iopub.execute_input":"2024-10-03T10:55:34.622960Z","iopub.status.idle":"2024-10-03T10:55:37.583002Z","shell.execute_reply.started":"2024-10-03T10:55:34.622721Z","shell.execute_reply":"2024-10-03T10:55:37.579988Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def f(x):\n tmp = x.copy()\n preprocess_input(tmp)\n return model(tmp)\n\nmasker_blur = shap.maskers.Image(\"blur(128,128)\", X[0].shape)\n\nexplainer_blur = shap.Explainer(f, masker_blur, output_names=class_names)\n\nshap_values_fine = explainer_blur(\n X[1:3], max_evals=500, batch_size=50, outputs=shap.Explanation.argsort.flip[:4]\n)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:55:37.586692Z","iopub.execute_input":"2024-10-03T10:55:37.587955Z","iopub.status.idle":"2024-10-03T10:57:22.689174Z","shell.execute_reply.started":"2024-10-03T10:55:37.587807Z","shell.execute_reply":"2024-10-03T10:57:22.687863Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# output with shap values\nshap.image_plot(shap_values_fine)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:57:22.691722Z","iopub.execute_input":"2024-10-03T10:57:22.692124Z","iopub.status.idle":"2024-10-03T10:57:23.888707Z","shell.execute_reply.started":"2024-10-03T10:57:22.692085Z","shell.execute_reply":"2024-10-03T10:57:23.887573Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"\n\n\n#SHAP for tabular data\n
","metadata":{}},{"cell_type":"code","source":"!pip install plotly\n!pip install xgboost","metadata":{"execution":{"iopub.status.busy":"2024-10-03T10:59:51.255931Z","iopub.execute_input":"2024-10-03T10:59:51.256937Z","iopub.status.idle":"2024-10-03T11:00:19.410582Z","shell.execute_reply.started":"2024-10-03T10:59:51.256891Z","shell.execute_reply":"2024-10-03T11:00:19.409036Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import matplotlib.pylab as pl\nimport numpy as np\nimport xgboost\nfrom sklearn.model_selection import train_test_split\nimport shap\n\n# print the JS visualization code to the notebook\nshap.initjs()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T11:01:00.371301Z","iopub.execute_input":"2024-10-03T11:01:00.371851Z","iopub.status.idle":"2024-10-03T11:01:00.727475Z","shell.execute_reply.started":"2024-10-03T11:01:00.371786Z","shell.execute_reply":"2024-10-03T11:01:00.726187Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"X, y = shap.datasets.adult()\nX_display, y_display = shap.datasets.adult(display=True)\n\nX_display.head()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T11:01:39.075670Z","iopub.execute_input":"2024-10-03T11:01:39.076156Z","iopub.status.idle":"2024-10-03T11:01:39.881005Z","shell.execute_reply.started":"2024-10-03T11:01:39.076112Z","shell.execute_reply":"2024-10-03T11:01:39.879670Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(y_display)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T11:01:51.240748Z","iopub.execute_input":"2024-10-03T11:01:51.241250Z","iopub.status.idle":"2024-10-03T11:01:51.362628Z","shell.execute_reply.started":"2024-10-03T11:01:51.241203Z","shell.execute_reply":"2024-10-03T11:01:51.360910Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# create a train/test split\nX_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=7)\nd_train = xgboost.DMatrix(X_train, label=y_train)\nd_test = xgboost.DMatrix(X_test, label=y_test)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T11:02:03.266082Z","iopub.execute_input":"2024-10-03T11:02:03.266542Z","iopub.status.idle":"2024-10-03T11:02:03.429294Z","shell.execute_reply.started":"2024-10-03T11:02:03.266500Z","shell.execute_reply":"2024-10-03T11:02:03.428171Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"params = {\n \"eta\": 0.01,\n \"objective\": \"binary:logistic\",\n \"subsample\": 0.5,\n \"base_score\": np.mean(y_train),\n \"eval_metric\": \"logloss\",\n}\nmodel = xgboost.train(\n params,\n d_train,\n 5000,\n evals=[(d_test, \"test\")],\n verbose_eval=100,\n early_stopping_rounds=20,\n)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T11:02:21.145696Z","iopub.execute_input":"2024-10-03T11:02:21.146569Z","iopub.status.idle":"2024-10-03T11:02:26.264108Z","shell.execute_reply.started":"2024-10-03T11:02:21.146522Z","shell.execute_reply":"2024-10-03T11:02:26.262898Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from sklearn.metrics import accuracy_score\n\n# Make predictions\ny_pred_prob = model.predict(d_test)\ny_pred = (y_pred_prob > 0.5).astype(int)\n\n# Calculate accuracy\naccuracy = accuracy_score(y_test, y_pred)\nprint(f\"Accuracy: {accuracy * 100:.2f}%\")","metadata":{"execution":{"iopub.status.busy":"2024-10-03T11:02:43.330456Z","iopub.execute_input":"2024-10-03T11:02:43.331252Z","iopub.status.idle":"2024-10-03T11:02:43.555411Z","shell.execute_reply.started":"2024-10-03T11:02:43.331209Z","shell.execute_reply":"2024-10-03T11:02:43.554287Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"xgboost.plot_importance(model,importance_type=\"weight\")\npl.title(\"xgboost.plot_importance(model)\")\npl.show()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T11:03:01.605950Z","iopub.execute_input":"2024-10-03T11:03:01.606779Z","iopub.status.idle":"2024-10-03T11:03:02.113781Z","shell.execute_reply.started":"2024-10-03T11:03:01.606731Z","shell.execute_reply":"2024-10-03T11:03:02.112440Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#The \"cover\" metric measures the coverage of a feature, which is the number of\n#samples or observations that are affected by splits involving that feature.\n#Essentially, it represents how frequently a feature is used to partition\n#the data and how many data points fall into those partitions.\n\nxgboost.plot_importance(model, importance_type=\"cover\")\npl.title('xgboost.plot_importance(model, importance_type=\"cover\")')\npl.show()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T11:03:14.120593Z","iopub.execute_input":"2024-10-03T11:03:14.121054Z","iopub.status.idle":"2024-10-03T11:03:14.553870Z","shell.execute_reply.started":"2024-10-03T11:03:14.121012Z","shell.execute_reply":"2024-10-03T11:03:14.552672Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#Gain represents the improvement in the objective function (such as accuracy or log loss)\n#that a feature provides when it is used in a split. It quantifies the\n#contribution of a feature to the model's performance.\n\nxgboost.plot_importance(model, importance_type=\"gain\")\npl.title('xgboost.plot_importance(model, importance_type=\"gain\")')\npl.show()","metadata":{"execution":{"iopub.status.busy":"2024-10-03T11:03:42.475731Z","iopub.execute_input":"2024-10-03T11:03:42.476214Z","iopub.status.idle":"2024-10-03T11:03:42.972143Z","shell.execute_reply.started":"2024-10-03T11:03:42.476171Z","shell.execute_reply":"2024-10-03T11:03:42.970907Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import plotly.io as pio\n# This takes 5-6 minutes since we are explaining over 30\n#thousand samples in a model with over a thousand trees\nexplainer = shap.TreeExplainer(model)\nshap_values = explainer.shap_values(X)\n","metadata":{"execution":{"iopub.status.busy":"2024-10-03T11:03:55.820468Z","iopub.execute_input":"2024-10-03T11:03:55.820920Z","iopub.status.idle":"2024-10-03T11:07:10.239377Z","shell.execute_reply.started":"2024-10-03T11:03:55.820876Z","shell.execute_reply":"2024-10-03T11:07:10.238279Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# print the JS visualization code to the notebook\nshap.initjs() # Ensure this line is uncommented\n\nshap.force_plot(explainer.expected_value, shap_values[100, :], X_display.iloc[100, :])","metadata":{"execution":{"iopub.status.busy":"2024-10-03T11:07:58.368721Z","iopub.execute_input":"2024-10-03T11:07:58.369231Z","iopub.status.idle":"2024-10-03T11:07:58.500031Z","shell.execute_reply.started":"2024-10-03T11:07:58.369186Z","shell.execute_reply":"2024-10-03T11:07:58.498709Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# print the JS visualization code to the notebook\nshap.initjs() # Ensure this line is uncommented\n\nshap.force_plot(\n explainer.expected_value, shap_values[:1000, :], X_display.iloc[:1000, :]\n)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T11:07:36.276082Z","iopub.execute_input":"2024-10-03T11:07:36.276508Z","iopub.status.idle":"2024-10-03T11:07:39.833199Z","shell.execute_reply.started":"2024-10-03T11:07:36.276469Z","shell.execute_reply":"2024-10-03T11:07:39.831906Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"shap.summary_plot(shap_values, X_display, plot_type=\"bar\")","metadata":{"execution":{"iopub.status.busy":"2024-10-03T11:08:13.308665Z","iopub.execute_input":"2024-10-03T11:08:13.309159Z","iopub.status.idle":"2024-10-03T11:08:13.703751Z","shell.execute_reply.started":"2024-10-03T11:08:13.309113Z","shell.execute_reply":"2024-10-03T11:08:13.702451Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"shap.summary_plot(shap_values, X)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T11:08:17.253397Z","iopub.execute_input":"2024-10-03T11:08:17.254516Z","iopub.status.idle":"2024-10-03T11:08:22.702588Z","shell.execute_reply.started":"2024-10-03T11:08:17.254465Z","shell.execute_reply":"2024-10-03T11:08:22.700813Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"for name in X_train.columns:\n shap.dependence_plot(name, shap_values, X, display_features=X_display)","metadata":{"execution":{"iopub.status.busy":"2024-10-03T11:08:23.528187Z","iopub.execute_input":"2024-10-03T11:08:23.529703Z","iopub.status.idle":"2024-10-03T11:08:35.353038Z","shell.execute_reply.started":"2024-10-03T11:08:23.529655Z","shell.execute_reply":"2024-10-03T11:08:35.351898Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"\n#Resources\n
\n\n#### SHAP: https://arxiv.org/abs/1602.04938\n#### SHAP GitHub: https://github.com/shap/shap\n#### SHAP docs: https://shap.readthedocs.io/en/latest/example_notebooks/tabular_examples/tree_based_models/Census%20income%20classification%20with%20XGBoost.html\n#### LIME: https://arxiv.org/abs/1602.04938\n#### LIME GitHub: https://github.com/marcotcr/lime\n#### YouTube: https://www.youtube.com/playlist?list=PLPTV0NXA_ZShaln9GfiHO_c0HzOSqLOGv","metadata":{}}]}