{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.10.13","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"},"kaggle":{"accelerator":"gpu","dataSources":[{"sourceId":8540,"databundleVersionId":862041,"sourceType":"competition"},{"sourceId":7799551,"sourceType":"datasetVersion","datasetId":4566712}],"dockerImageVersionId":30665,"isInternetEnabled":true,"language":"python","sourceType":"notebook","isGpuEnabled":true}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"code","source":"# Imort packages\nimport pandas as pd\nimport numpy as np\nimport matplotlib.pyplot as plt\nimport seaborn as sns\nfrom sklearn.ensemble import GradientBoostingClassifier\nfrom sklearn.svm import SVC\nfrom sklearn.linear_model import LogisticRegression\nfrom tqdm import tqdm\nfrom imblearn.over_sampling import SMOTE\nfrom sklearn.utils import resample\nfrom sklearn.model_selection import cross_val_score\nimport xgboost as xgb\nimport lightgbm as lgb\nfrom sklearn.ensemble import RandomForestClassifier\nfrom sklearn.ensemble import AdaBoostClassifier\nfrom sklearn.tree import DecisionTreeClassifier\nfrom sklearn.svm import LinearSVC\nfrom sklearn.calibration import CalibratedClassifierCV\nfrom sklearn.ensemble import StackingClassifier\nfrom scipy.stats import mode\nfrom xgboost import XGBClassifier\nfrom hyperopt import space_eval\nfrom sklearn.feature_extraction.text import CountVectorizer\nfrom sklearn.decomposition import LatentDirichletAllocation\nfrom tqdm import tqdm\nimport pandas as pd\n\nfrom sklearn.preprocessing import StandardScaler\nfrom sklearn.pipeline import make_pipeline\nfrom sklearn.model_selection import train_test_split, RandomizedSearchCV\nfrom sklearn.metrics import accuracy_score\nfrom sklearn.metrics import roc_auc_score\nimport warnings\nwarnings.filterwarnings(\"ignore\")","metadata":{"_uuid":"8fb484df-4355-412a-9a8f-3064e7106ade","_cell_guid":"e7e01c6a-0cfa-4ad2-b019-c1c8ae48bf39","collapsed":false,"jupyter":{"outputs_hidden":false},"execution":{"iopub.status.busy":"2024-03-24T02:04:11.490589Z","iopub.execute_input":"2024-03-24T02:04:11.490925Z","iopub.status.idle":"2024-03-24T02:04:11.500407Z","shell.execute_reply.started":"2024-03-24T02:04:11.490897Z","shell.execute_reply":"2024-03-24T02:04:11.499517Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Data Cleaning","metadata":{}},{"cell_type":"code","source":"train_original = pd.read_csv('/kaggle/input/click-fraud/train_resampled.csv')","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:04:11.501792Z","iopub.execute_input":"2024-03-24T02:04:11.502061Z","iopub.status.idle":"2024-03-24T02:04:11.738122Z","shell.execute_reply.started":"2024-03-24T02:04:11.502033Z","shell.execute_reply":"2024-03-24T02:04:11.737113Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Data Overview\n\n* ip: ip address of click\n* app: app id for marketing\n* device: device type id of user mobile phone\n* os: os version id of user mobile phone\n* channel: channel id of mobile ad publisher(it is the id of website on which the ads were published)\n* click_time: time stamp of click (UTC)\n* is_attributed: the target that is to be predicted, indicating the app was downloaded\n\n\nPerformance Metric\n* AUC score\n\nAll the features(ip, app, device ,os and channel) are encoded. I will be handling them as categorical later.","metadata":{}},{"cell_type":"code","source":"train = train_original.copy()","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:05:36.496959Z","iopub.execute_input":"2024-03-24T02:05:36.497326Z","iopub.status.idle":"2024-03-24T02:05:36.504562Z","shell.execute_reply.started":"2024-03-24T02:05:36.497297Z","shell.execute_reply":"2024-03-24T02:05:36.503659Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train.drop(labels='Unnamed: 0', axis=1, inplace=True)","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:05:39.920152Z","iopub.execute_input":"2024-03-24T02:05:39.920510Z","iopub.status.idle":"2024-03-24T02:05:39.942128Z","shell.execute_reply.started":"2024-03-24T02:05:39.920480Z","shell.execute_reply":"2024-03-24T02:05:39.941257Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def transformation(data):\n    data['click_time'] = pd.to_datetime(data.click_time)\n    data['day'] = data['click_time'].dt.day.astype('uint8')\n    data['hour'] = data['click_time'].dt.hour.astype('uint8')\n    data['minute'] = data['click_time'].dt.minute.astype('uint8')\n    data['second'] = data['click_time'].dt.second.astype('uint8')\n    #df.drop(labels='click_time', axis=1, inplace = True)\n    return data","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:05:43.610220Z","iopub.execute_input":"2024-03-24T02:05:43.610952Z","iopub.status.idle":"2024-03-24T02:05:43.616772Z","shell.execute_reply.started":"2024-03-24T02:05:43.610916Z","shell.execute_reply":"2024-03-24T02:05:43.615686Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# transformation\ntrain = transformation(train)","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:05:45.060488Z","iopub.execute_input":"2024-03-24T02:05:45.061206Z","iopub.status.idle":"2024-03-24T02:05:45.120372Z","shell.execute_reply.started":"2024-03-24T02:05:45.061173Z","shell.execute_reply":"2024-03-24T02:05:45.119592Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Feature Engineering","metadata":{}},{"cell_type":"markdown","source":"## Statistical features","metadata":{}},{"cell_type":"code","source":"#Funtion to extract group features\ndef extractGroupFeat(df,group_list,select,agg_func,type):\n    \"\"\" This function extracts aggregate features based on group provided \"\"\"\n    use_col = group_list + [select]\n    feat_name = '_'.join(group_list) + '_' + agg_func+ '_' +'on' + '_' + select\n    feat = df[use_col].groupby(group_list)[select].agg(agg_func).reset_index()\n    feat = feat.rename(columns = {select:feat_name})\n    df = df.merge(feat, on=group_list, how='left')\n    df[feat_name] = df[feat_name].astype(type)\n    return df\n\n#Function to extract cummulative count feature\ndef extractCumFeat(df,group_list,select,type):\n    \"\"\" This function extratcts cummulative counts of a feature based on group provided \"\"\"\n    use_col = group_list + [select]\n    feat_name = '_'.join(group_list) + '_' + 'cum' + '_' +'on' + '_' + 'select'\n    feat =  df[use_col].groupby(group_list)[select].agg('cumcount')\n    df[feat_name] = feat\n    df[feat_name] = df[feat_name].astype(type)\n    return df","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:05:46.914883Z","iopub.execute_input":"2024-03-24T02:05:46.915231Z","iopub.status.idle":"2024-03-24T02:05:46.924150Z","shell.execute_reply.started":"2024-03-24T02:05:46.915202Z","shell.execute_reply":"2024-03-24T02:05:46.923088Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Number of unique apps per ip, device and os\ntrain = extractGroupFeat(train, ['ip', 'device', 'os'], 'app', 'nunique', 'int64')\n# Number of unique click hour for each ip every day\ntrain = extractGroupFeat(train, ['ip', 'day'], 'hour', 'nunique', 'int64')\n# Number of unique apps per ip\ntrain = extractGroupFeat(train, ['ip'], 'app', 'nunique', 'int64')\n# Cumulative count for os for ip\ntrain = extractCumFeat(train, ['ip'], 'os', 'int64')\n# Cumulative count of apps per ip, device, and os\ntrain = extractCumFeat(train, ['ip', 'device', 'os'], 'app', 'int64')\n# Total count for ip, app, day and hour combination\ntrain = extractGroupFeat(train, ['ip', 'app', 'day', 'hour'], 'click_time', 'count', 'int64')\n# Total count for ip, app combination\ntrain = extractGroupFeat(train, ['ip', 'app'], 'click_time', 'count', 'int64')\n# Total count for ip, app, os combination\ntrain = extractGroupFeat(train, ['ip', 'app', 'os'], 'click_time', 'count', 'int64')\n# Variance in hour for ip, app, channel\ntrain = extractGroupFeat(train, ['ip', 'app', 'os'], 'hour', 'var', 'float32')\n# Mean hour for ip, app, channel\ntrain = extractGroupFeat(train, ['ip', 'app', 'channel'], 'hour', 'mean', 'float32')\n# Average click on app by distinct user\nuser_clicks = train.groupby(['app', 'ip']).size().reset_index(name='user_clicks')\navg_clicks = user_clicks.groupby('app')['user_clicks'].mean().reset_index(name='avg_clicks_per_user')\ntrain = train.merge(avg_clicks, on='app', how='left')\n# Number of unique os per ip and app\ntrain = extractGroupFeat(train, ['ip', 'app'], 'os', 'nunique', 'int64')\n# Number of unique device per ip\ntrain = extractGroupFeat(train, ['ip'], 'device', 'nunique', 'int64')\n# Number of unique channels per app\ntrain = extractGroupFeat(train, ['app'], 'channel', 'nunique', 'int64')\n# Number of unique channels per ip\ntrain = extractGroupFeat(train, ['ip'], 'channel', 'nunique', 'int64')\n# Variance in day for ip, app and os\ntrain = extractGroupFeat(train, ['ip','app','os'], 'channel', 'var', 'float32')\n# Total click count for ip, device, os\ntrain = extractGroupFeat(train, ['ip', 'device', 'os'], 'click_time', 'count', 'int64')\n# Total click count for ip\ntrain = extractGroupFeat(train, ['ip'], 'click_time', 'count', 'int64')\n# Variance in hour for ip\ntrain = extractGroupFeat(train, ['ip'], 'hour', 'var', 'float32')","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:05:48.563067Z","iopub.execute_input":"2024-03-24T02:05:48.563711Z","iopub.status.idle":"2024-03-24T02:05:49.468737Z","shell.execute_reply.started":"2024-03-24T02:05:48.563675Z","shell.execute_reply":"2024-03-24T02:05:49.467941Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"I developed these features to distinguish statistical differences between scenarios. For instance, a low total unique app count for a given IP, device, and OS combination suggests repetitive clicking on the same apps without downloading them. The reasoning is that once an app is downloaded, the likelihood of clicking on it diminishes, whereas the probability of exploring and clicking on different apps increases, resulting in a higher count of unique apps.","metadata":{}},{"cell_type":"markdown","source":"Next I created time related features. For this I used click time feature and extracted new click delta features(difference in time between to clicks) by grouping existing categorical features in various combinations.","metadata":{}},{"cell_type":"code","source":"#Function to extract next click on the app based on group provided\ndef extractNextClick(df,group_list,k):\n    \"\"\" This function extract diff between current and next click based on group provided \"\"\"\n    use_col = group_list + ['click_time']\n    feat_name = '_'.join(group_list) + '_' + 'next_click' + str(k)\n    df[feat_name] = (df[use_col].groupby(group_list)['click_time'].shift(k) - df['click_time']).dt.seconds.astype('float32')\n    return df\n\n#Function to extract prev click on the app based on group provided\ndef extractPrevClick(df,group_list,k):\n    \"\"\" This function extract diff between current and prev click based on group provided \"\"\"\n    use_col = group_list + ['click_time']\n    feat_name = '_'.join(group_list) + '_' + 'prev_click' + str(k)\n    df[feat_name] = (df['click_time'] - df[use_col].groupby(group_list)['click_time'].shift(k)).dt.seconds.astype('float32')\n    return df","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:05:50.962366Z","iopub.execute_input":"2024-03-24T02:05:50.962787Z","iopub.status.idle":"2024-03-24T02:05:50.970292Z","shell.execute_reply.started":"2024-03-24T02:05:50.962747Z","shell.execute_reply":"2024-03-24T02:05:50.969432Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train = extractNextClick(train, ['ip', 'os', 'device'], 1)\ntrain = extractPrevClick(train, ['ip', 'channel'], 1)\ntrain = extractNextClick(train, ['ip', 'app', 'channel', 'device', 'os'], 2)\ntrain = extractPrevClick(train, ['ip', 'os'], 1)","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:05:52.561148Z","iopub.execute_input":"2024-03-24T02:05:52.561481Z","iopub.status.idle":"2024-03-24T02:05:52.670519Z","shell.execute_reply.started":"2024-03-24T02:05:52.561457Z","shell.execute_reply":"2024-03-24T02:05:52.669751Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"These feature can be interpreted as if the time difference between current click and next click or previous click is very less then there is a high chance the click is fraud as if the previous click lead to download then there is no point of clicking on the app ad again and again.","metadata":{}},{"cell_type":"markdown","source":"Then I created ratio features","metadata":{}},{"cell_type":"code","source":"#Function to get top kth ratio counts\ndef getTopCountRatio(df, group_list, select, k, type):\n    \"\"\" This function calculates top kth ratio counts\"\"\"\n    use_col = group_list + [select]\n    feat_name = '_'.join(group_list) + '_' + str(k) + 'count_ratio' + '_' +'on' + '_' + select\n    feat = df[use_col].groupby(group_list)[select].agg(\n        lambda x: float(list(x.value_counts())[k]/len(x)) if len(list(x.value_counts())) > k else 0\n    ).reset_index()\n    feat = feat.rename(columns = {select:feat_name}) \n    df = df.merge(feat, on=group_list, how='left')\n    df[feat_name] = df[feat_name].astype(type)\n    return df\n\n#Function to get nunique to count ratio\ndef getNCountRatio(df,group_list,select,type):\n    \"\"\" This function calculates nunique to count ratio \"\"\"\n    use_col = group_list + [select]\n    feat_name = '_'.join(group_list) + '_' + 'nunique_count_ratio' + '_' +'on' + '_' + select\n    feat = df[use_col].groupby(group_list)[select].agg(lambda x: float(len(x.unique())/len(x))).reset_index()\n    feat = feat.rename(columns = {select:feat_name})\n    df = df.merge(feat, on=group_list, how='left')\n    df[feat_name] = df[feat_name].astype(type)\n    return df","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:05:53.986497Z","iopub.execute_input":"2024-03-24T02:05:53.987184Z","iopub.status.idle":"2024-03-24T02:05:53.997706Z","shell.execute_reply.started":"2024-03-24T02:05:53.987151Z","shell.execute_reply":"2024-03-24T02:05:53.996673Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Top device count ratio for ip\ntrain = getTopCountRatio(train, ['ip'], 'device', 0, 'float32')\n# Unique count ratio of app for every ip\ntrain = getNCountRatio(train, ['ip'], 'app', 'float32')\n# Unique count ratio of hour for every ip\ntrain = getNCountRatio(train, ['ip'], 'hour', 'float32')\n# Unique count ratio for app for every ip,os,device\ntrain = getNCountRatio(train, ['ip', 'os', 'device'], 'app', 'float32')\n# Top app count ratio for channel\ntrain = getTopCountRatio(train, ['channel'], 'app', 0, 'float32')\n# Top channel count ratio for app\ntrain = getTopCountRatio(train, ['app'], 'channel', 0, 'float32')\n# Top app count ratio for ip\ntrain = getTopCountRatio(train, ['ip'], 'app', 0, 'float32')\n# Unique count ratio for channel for every day,ip\ntrain = getNCountRatio(train, ['day', 'ip'], 'channel', 'float32')","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:05:55.884457Z","iopub.execute_input":"2024-03-24T02:05:55.885082Z","iopub.status.idle":"2024-03-24T02:06:42.831148Z","shell.execute_reply.started":"2024-03-24T02:05:55.885049Z","shell.execute_reply":"2024-03-24T02:06:42.830144Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Top feature count is the ratio of top value count for the feature is to total number of values for the given combination. For example if this ratio for app is calculated on ip and it is high then it shows that a particular app is being clicked by this ip most of the time as compared to other apps clicked by the same ip.","metadata":{"execution":{"iopub.status.busy":"2024-03-23T13:52:20.043405Z","iopub.execute_input":"2024-03-23T13:52:20.043738Z","iopub.status.idle":"2024-03-23T13:52:20.048663Z","shell.execute_reply.started":"2024-03-23T13:52:20.043710Z","shell.execute_reply":"2024-03-23T13:52:20.047691Z"}}},{"cell_type":"markdown","source":"Top unique count ratio is the ratio of number of unique values of the feature is to total number of values for the feature for a given combination. Now if the ratio is high then there are many unique values for the given feature occuring for the given combination.","metadata":{}},{"cell_type":"markdown","source":"## Latent Dirichlet Allocation(LDA)","metadata":{}},{"cell_type":"markdown","source":"LDA is a topic modelling technique(extracting the hidden topics from large volumes of text). It states that each document is combination of multiple topics and each word in the document contributes to one of those topics.\n\nHere I used lda to create 5 topics features for every ip using the apps.","metadata":{}},{"cell_type":"code","source":"#Function to create topic probablities \ndef modelTopics(df,col1,col2,type):\n    \"\"\" This function creates topic probablities for given column\"\"\"\n    vocab = list(dict(df[col2].value_counts()[:10]).keys())\n    col2_of_col1 = {}\n    for i in tqdm(range(len(df))):\n        col2_of_col1.setdefault(df.loc[i,col1], [])\n        if(df.loc[i,col2] in vocab):\n            col2_of_col1[df.loc[i,col1]].append(str(df.loc[i,col2]))\n    lcol1 = list(col2_of_col1.keys())\n    sentences = [' '.join(col2_of_col1[col]) for col in lcol1]\n    cv = CountVectorizer()\n    as_matrix = cv.fit_transform(sentences)\n    LDA = LatentDirichletAllocation(n_components=5)\n    topics = LDA.fit_transform(as_matrix)\n    col1_col2_topic = {}\n    for i in tqdm(range(len(lcol1))):\n        col1_col2_topic[lcol1[i]] = topics[i]\n    col_names = {'index':col1,0:col1+col2+'topic1',1:col1+col2+'topic2',2:col1+col2+'topic3',3:col1+col2+'topic4', 4:col1+col2+'topic5'}\n    topic_df = pd.DataFrame.from_dict(col1_col2_topic, orient='index').reset_index().rename(columns= col_names)\n    df = df.merge(topic_df, on=['ip'], how='left')\n    df[col1+col2+'topic1'] = df[col1+col2+'topic1'].astype(type)\n    df[col1+col2+'topic2'] = df[col1+col2+'topic2'].astype(type)\n    df[col1+col2+'topic3'] = df[col1+col2+'topic3'].astype(type)\n    df[col1+col2+'topic4'] = df[col1+col2+'topic4'].astype(type)\n    df[col1+col2+'topic5'] = df[col1+col2+'topic5'].astype(type)\n    return df ","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:06:42.832770Z","iopub.execute_input":"2024-03-24T02:06:42.833061Z","iopub.status.idle":"2024-03-24T02:06:42.844453Z","shell.execute_reply.started":"2024-03-24T02:06:42.833036Z","shell.execute_reply":"2024-03-24T02:06:42.843620Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train = modelTopics(train, 'ip', 'app', 'float32')","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:06:42.845616Z","iopub.execute_input":"2024-03-24T02:06:42.845893Z","iopub.status.idle":"2024-03-24T02:07:17.942109Z","shell.execute_reply.started":"2024-03-24T02:06:42.845871Z","shell.execute_reply":"2024-03-24T02:07:17.940988Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Now, from the existing features I kept only app, device and os. Ip can be allocated to any user and the unique values for ip is also high so I discarded it. Channel was causing overfitting so I discarded it as well.","metadata":{}},{"cell_type":"code","source":"# delete features\ntrain = train.drop(columns=['ip','channel','attributed_time'])","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:07:17.944376Z","iopub.execute_input":"2024-03-24T02:07:17.944703Z","iopub.status.idle":"2024-03-24T02:07:17.957561Z","shell.execute_reply.started":"2024-03-24T02:07:17.944673Z","shell.execute_reply":"2024-03-24T02:07:17.956477Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Splitting dataset","metadata":{}},{"cell_type":"markdown","source":"I used time based splitting to split the data set into two parts train and cross validation with 80 and 20% of data respectively.","metadata":{}},{"cell_type":"code","source":"# Sort the DataFrame by 'click_time'\ntrain_sorted = train.sort_values(by='click_time')\n\n# Calculate the split point (80% of the dataset)\nsplit_point = int(len(train_sorted) * 0.8)\n\n# Split the data into training and cross-validation sets\ntrain_set = train_sorted.iloc[:split_point]\ncv_set = train_sorted.iloc[split_point:]","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:07:17.959312Z","iopub.execute_input":"2024-03-24T02:07:17.960077Z","iopub.status.idle":"2024-03-24T02:07:17.999576Z","shell.execute_reply.started":"2024-03-24T02:07:17.960040Z","shell.execute_reply":"2024-03-24T02:07:17.998614Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_set.drop(columns = ['click_time'], inplace=True)\ncv_set.drop(columns = ['click_time'], inplace=True)","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:07:18.000829Z","iopub.execute_input":"2024-03-24T02:07:18.001184Z","iopub.status.idle":"2024-03-24T02:07:18.020369Z","shell.execute_reply.started":"2024-03-24T02:07:18.001158Z","shell.execute_reply":"2024-03-24T02:07:18.019446Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Modeling","metadata":{}},{"cell_type":"markdown","source":"## Light GBM","metadata":{}},{"cell_type":"code","source":"from hyperopt import hp, fmin, tpe, STATUS_OK, Trials\n\n# Define the space for hyperparameters\nspace = {\n    'objective': 'binary',\n    'metric': 'auc',\n    'boosting': 'gbdt',\n    'is_unbalance': True,\n    'n_estimators': hp.choice('n_estimators', [100, 250, 500, 750]),\n    'max_depth': hp.choice('max_depth', [5, 10, 15, 20]),\n    'num_leaves': hp.choice('num_leaves', range(2, 2**5 + 1)),\n    'subsample': hp.uniform('subsample', 0.5, 1),\n    'colsample_bytree': hp.uniform('colsample_bytree', 0.5, 1),\n    'learning_rate': hp.uniform('learning_rate', 0.01, 0.2),\n    'reg_alpha': hp.uniform('reg_alpha', 0, 1),\n    'reg_lambda': hp.uniform('reg_lambda', 0, 1),\n    'min_child_samples': hp.choice('min_child_samples', [20, 50, 100]),\n    'num_boost_round': 100\n}\n\n# setting columns and params\nuse_col = ['app', 'device', 'os','ip_device_os_nunique_on_app',\n       'ip_day_nunique_on_hour', 'ip_nunique_on_app', 'ip_cum_on_select',\n       'ip_device_os_cum_on_select', 'ip_app_day_hour_count_on_click_time',\n       'ip_app_count_on_click_time', 'ip_app_os_count_on_click_time',\n       'ip_app_os_var_on_hour', 'ip_app_channel_mean_on_hour',\n       'avg_clicks_per_user', 'ip_app_nunique_on_os', 'ip_nunique_on_device',\n       'app_nunique_on_channel', 'ip_nunique_on_channel',\n       'ip_app_os_var_on_channel', 'ip_device_os_count_on_click_time',\n       'ip_count_on_click_time', 'ip_var_on_hour', 'ip_os_device_next_click1',\n       'ip_channel_prev_click1', 'ip_app_channel_device_os_next_click2',\n       'ip_os_prev_click1', 'ip_0count_ratio_on_device',\n       'ip_nunique_count_ratio_on_app', 'ip_nunique_count_ratio_on_hour',\n       'ip_os_device_nunique_count_ratio_on_app',\n       'channel_0count_ratio_on_app', 'app_0count_ratio_on_channel',\n       'ip_0count_ratio_on_app', 'day_ip_nunique_count_ratio_on_channel',\n       'ipapptopic1', 'ipapptopic2', 'ipapptopic3', 'ipapptopic4',\n       'ipapptopic5']\n\ntarget = 'is_attributed'\n\ncategorical = ['app', 'device','os']","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:07:18.021657Z","iopub.execute_input":"2024-03-24T02:07:18.022573Z","iopub.status.idle":"2024-03-24T02:07:18.034665Z","shell.execute_reply.started":"2024-03-24T02:07:18.022535Z","shell.execute_reply":"2024-03-24T02:07:18.033685Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def hyperparameter_tuning(space):\n    xgtrain = lgb.Dataset(train_set[use_col].values, label=train_set[target].values,feature_name=use_col,categorical_feature=categorical, params={'verbose': -1})\n    xgvalid = lgb.Dataset(cv_set[use_col].values, label=cv_set[target].values,feature_name=use_col,categorical_feature=categorical, params={'verbose': -1})\n    params={'objective': space['objective'] ,\n               'metric':space['metric'],\n               'boosting':space['boosting'],\n               'is_unbalance': space['is_unbalance'],\n               'n_estimators':space['n_estimators'],\n               'max_depth': space['max_depth'] ,\n               'num_leaves':space['num_leaves'] ,\n               'subsample':space['subsample'] ,\n               'colsample_bytree': space['colsample_bytree'] ,\n               'learning_rate':space['learning_rate'] ,\n               'reg_alpha': space['reg_alpha'],\n               'reg_lambda': space['reg_lambda'] ,\n               'min_child_samples': space['min_child_samples'] ,\n               'force_col_wise': 'true'\n            \n               \n    }\n    model = lgb.train(params, xgtrain, valid_sets=[xgtrain, xgvalid], valid_names=['train','valid'],num_boost_round=space['num_boost_round'])\n    pred = model.predict(cv_set[use_col])\n    auc = roc_auc_score(cv_set[target], pred)\n    print (\"SCORE:\", auc)\n    return {'loss': -auc, 'status': STATUS_OK }\n\nprint(\"Finding best params....\")\ntrials = Trials()\nbest = fmin(fn=hyperparameter_tuning,\n            space=space,\n            algo=tpe.suggest,\n            max_evals=20,\n            trials=trials)\n\nprint (best)","metadata":{"execution":{"iopub.status.busy":"2024-03-23T09:22:30.903528Z","iopub.execute_input":"2024-03-23T09:22:30.903915Z","iopub.status.idle":"2024-03-23T09:24:33.474919Z","shell.execute_reply.started":"2024-03-23T09:22:30.903887Z","shell.execute_reply":"2024-03-23T09:24:33.473984Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Set up the best parameters from hyperopt output\nbest_params = {\n    'colsample_bytree': 0.6318812714994927,\n    'learning_rate': 0.08445169672164289,\n    'max_depth': 13 + 5,  \n    'min_child_samples': 2, \n    'n_estimators': 3 + 100,  \n    'num_leaves': 14,\n    'reg_alpha': 0.37747740442925026,\n    'reg_lambda': 0.6987700507748307,\n    'subsample': 0.7389179714984548,\n    'objective': 'binary',\n    'metric': 'auc',\n    'boosting': 'gbdt',\n    'is_unbalance': True,\n    'force_row_wise': True\n}\n\n# Train the model on the training set\nprint(\"creating train and val sets...\")\ntrain_data = lgb.Dataset(train_set[use_col], label=train_set[target], categorical_feature=categorical, free_raw_data=False)\ncv_data = lgb.Dataset(cv_set[use_col], label=cv_set[target], reference=train_data, free_raw_data=False)\n\nprint(\"training model...\")\nfinal_model = lgb.train(best_params, train_data, valid_sets=[train_data, cv_data], valid_names=['train', 'valid'], num_boost_round=100,\n                       callbacks=[lgb.early_stopping(stopping_rounds=50), lgb.log_evaluation(50)])\n\n# Predict on the cross-validation set and calculate AUC\npredictions = final_model.predict(cv_set[use_col])\nauc_score = roc_auc_score(cv_set[target], predictions)\nprint(f\"AUC Score: {auc_score}\")\n\n# Plot feature importance\nlgb.plot_importance(final_model, max_num_features=30, importance_type='split')\nplt.title(\"LGBM Feature Importance\")\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:07:18.035853Z","iopub.execute_input":"2024-03-24T02:07:18.036172Z","iopub.status.idle":"2024-03-24T02:07:25.513434Z","shell.execute_reply.started":"2024-03-24T02:07:18.036140Z","shell.execute_reply":"2024-03-24T02:07:25.512419Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Neuron Network","metadata":{}},{"cell_type":"markdown","source":"LightGBM handles missing values and categorical values by specifying the categorical features but for neural network we need to handle these two conditions.\n\nHandling missing values","metadata":{}},{"cell_type":"code","source":"# Checking for columns in the DataFrame that have null values\nnull_columns = train.columns[train.isnull().any()]\n\n# Printing column names and the count of null values in each column\nfor column in null_columns:\n    print(f\"{column}: {train[column].isnull().sum()} null values\")","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:07:25.514733Z","iopub.execute_input":"2024-03-24T02:07:25.515036Z","iopub.status.idle":"2024-03-24T02:07:25.528531Z","shell.execute_reply.started":"2024-03-24T02:07:25.515010Z","shell.execute_reply":"2024-03-24T02:07:25.527560Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"I used maximum value for filling missing values in time delta features because if no value is present then the previous or next click has not happened yet so max value is more suitable compared to mean. For rest of the features I used mean value to fill the missing values.","metadata":{}},{"cell_type":"code","source":"fill_max = ['ip_os_prev_click1', 'ip_app_channel_device_os_next_click2', 'ip_channel_prev_click1',\n           'ip_os_device_next_click1']\n\nfill_mean = ['ip_app_os_var_on_hour', 'ip_app_os_var_on_channel', 'ip_var_on_hour']","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:07:25.532617Z","iopub.execute_input":"2024-03-24T02:07:25.532973Z","iopub.status.idle":"2024-03-24T02:07:25.538257Z","shell.execute_reply.started":"2024-03-24T02:07:25.532946Z","shell.execute_reply":"2024-03-24T02:07:25.537180Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# filling missing values with max\nfor col in fill_max:\n    train[col].fillna(train[col].max(), inplace=True)\n \n# filling miss values with mean\nfor col in fill_mean:\n    train[col].fillna(train[col].mean(), inplace=True)","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:07:25.539848Z","iopub.execute_input":"2024-03-24T02:07:25.540152Z","iopub.status.idle":"2024-03-24T02:07:25.568982Z","shell.execute_reply.started":"2024-03-24T02:07:25.540128Z","shell.execute_reply":"2024-03-24T02:07:25.567907Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Handling categorical features -\n\nThere number of unique values for every categorical features is very large and one hot encoding these features will end up creating a large number of features. So for encoding these categorical features I used response encoding.\n\nResponse encoding is a technique of encoding the categorical features using the target variables. In response encoding every categorical feature is converted into features equal to the number of different classes present in the target variable and the encoding for each value for a class is the probability of occurence of that value with respect to that class label.\n\nBelow figure summarizes the response encoding-Handling categorical features -\n\nThere number of unique values for every categorical features is very large and one hot encoding these features will end up creating a large number of features. So for encoding these categorical features I used response encoding.\n\nResponse encoding is a technique of encoding the categorical features using the target variables. In response encoding every categorical feature is converted into features equal to the number of different classes present in the target variable and the encoding for each value for a class is the probability of occurence of that value with respect to that class label.\n\nBelow figure summarizes the response encoding!\n\n![image.png](attachment:e34f8a5d-65b6-45e9-ae75-d1b883334713.png)!","metadata":{},"attachments":{"e34f8a5d-65b6-45e9-ae75-d1b883334713.png":{"image/png":"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"}}},{"cell_type":"code","source":"# Function for response encoding\ndef responseEncode(df,col):\n    \"\"\"This function encodes categorical variable based on response\"\"\"\n    featName1 = col+'_class1'\n    featName2 = col+'_class0'\n    g = df[[col,'is_attributed']].groupby([col])['is_attributed'].agg(['count','sum']).reset_index().rename(columns={'sum':'class1'})\n    g['class0'] = g['count'] - g['class1']\n    g[featName1] = g['class1']/g['count']\n    g[featName2] = g['class0']/g['count']\n    \n    return g.drop(['count','class0','class1'],axis=1)","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:07:25.570470Z","iopub.execute_input":"2024-03-24T02:07:25.570910Z","iopub.status.idle":"2024-03-24T02:07:25.582403Z","shell.execute_reply.started":"2024-03-24T02:07:25.570872Z","shell.execute_reply":"2024-03-24T02:07:25.581091Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Total number of new features created by response encoding is equal to #categorical_features * #number of unique_class_labels.","metadata":{}},{"cell_type":"code","source":"categorical_features = ['app', 'device', 'os']\n\n# Assuming 'train' is your main DataFrame\nfor col in categorical_features:\n    encoded_df = responseEncode(train, col)  # Encode the feature\n    \n    # Merge the encoded features back into the original DataFrame\n    train = train.merge(encoded_df, on=col, how='left')","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:07:25.585839Z","iopub.execute_input":"2024-03-24T02:07:25.586131Z","iopub.status.idle":"2024-03-24T02:07:25.680406Z","shell.execute_reply.started":"2024-03-24T02:07:25.586105Z","shell.execute_reply":"2024-03-24T02:07:25.679275Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Before training neural network we also need to scale numerical columns. I used min max scaling for scaling numerical features.","metadata":{}},{"cell_type":"code","source":"from sklearn.preprocessing import MinMaxScaler\n\n\nnumerical_features = ['ip_device_os_nunique_on_app',\n       'ip_day_nunique_on_hour', 'ip_nunique_on_app', 'ip_cum_on_select',\n       'ip_device_os_cum_on_select', 'ip_app_day_hour_count_on_click_time',\n       'ip_app_count_on_click_time', 'ip_app_os_count_on_click_time',\n       'ip_app_os_var_on_hour', 'ip_app_channel_mean_on_hour',\n       'avg_clicks_per_user', 'ip_app_nunique_on_os', 'ip_nunique_on_device',\n       'app_nunique_on_channel', 'ip_nunique_on_channel',\n       'ip_app_os_var_on_channel', 'ip_device_os_count_on_click_time',\n       'ip_count_on_click_time', 'ip_var_on_hour', 'ip_os_device_next_click1',\n       'ip_channel_prev_click1', 'ip_app_channel_device_os_next_click2',\n       'ip_os_prev_click1', 'ip_0count_ratio_on_device',\n       'ip_nunique_count_ratio_on_app', 'ip_nunique_count_ratio_on_hour',\n       'ip_os_device_nunique_count_ratio_on_app',\n       'channel_0count_ratio_on_app', 'app_0count_ratio_on_channel',\n       'ip_0count_ratio_on_app', 'day_ip_nunique_count_ratio_on_channel',\n       'ipapptopic1', 'ipapptopic2', 'ipapptopic3', 'ipapptopic4',\n       'ipapptopic5'] \n\n# Initialize the scaler\nscaler = MinMaxScaler()\n\n# Fit the scaler to your data and transform the numerical features\ntrain[numerical_features] = scaler.fit_transform(train[numerical_features])","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:07:25.681764Z","iopub.execute_input":"2024-03-24T02:07:25.682086Z","iopub.status.idle":"2024-03-24T02:07:25.748695Z","shell.execute_reply.started":"2024-03-24T02:07:25.682057Z","shell.execute_reply":"2024-03-24T02:07:25.747682Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import tensorflow as tf\nfrom tensorflow import keras\nfrom tensorflow.keras import layers, models\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.metrics import roc_auc_score\nfrom tensorflow.keras.callbacks import EarlyStopping","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:07:25.749997Z","iopub.execute_input":"2024-03-24T02:07:25.750395Z","iopub.status.idle":"2024-03-24T02:07:37.396455Z","shell.execute_reply.started":"2024-03-24T02:07:25.750369Z","shell.execute_reply":"2024-03-24T02:07:37.395661Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"X = train.drop(columns = ['is_attributed', 'click_time', 'app', 'device', 'os'])\ny = train['is_attributed']","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:07:37.397517Z","iopub.execute_input":"2024-03-24T02:07:37.398240Z","iopub.status.idle":"2024-03-24T02:07:37.428210Z","shell.execute_reply.started":"2024-03-24T02:07:37.398214Z","shell.execute_reply":"2024-03-24T02:07:37.427275Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.25, stratify=y, random_state=43)","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:07:37.429277Z","iopub.execute_input":"2024-03-24T02:07:37.429546Z","iopub.status.idle":"2024-03-24T02:07:37.554568Z","shell.execute_reply.started":"2024-03-24T02:07:37.429521Z","shell.execute_reply":"2024-03-24T02:07:37.553466Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Adjust the model architecture\nmodel = keras.Sequential([\n    layers.Input(shape=(X_train.shape[1],)),\n    layers.Dense(256, activation='relu'),\n    layers.Dropout(0.3),\n    layers.Dense(128, activation='relu'),\n    layers.Dropout(0.3),\n    layers.BatchNormalization(),\n    layers.Dense(64, activation='relu'),\n    layers.Dense(1, activation='sigmoid')\n])\n\n# Compile the model\nmodel.compile(optimizer='adam',\n              loss='binary_crossentropy',\n              metrics=[tf.keras.metrics.AUC(name='auc')])\n\n# Add EarlyStopping\nearly_stopping = EarlyStopping(monitor='val_auc', patience=10, mode='max', restore_best_weights=True)\n\n# Fit the model\nhistory = model.fit(X_train, y_train,\n                    validation_data=(X_test, y_test),\n                    epochs=50,\n                    batch_size=128,\n                    callbacks=[early_stopping],\n                    verbose=1)\n","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:07:37.555831Z","iopub.execute_input":"2024-03-24T02:07:37.556112Z","iopub.status.idle":"2024-03-24T02:09:00.188246Z","shell.execute_reply.started":"2024-03-24T02:07:37.556088Z","shell.execute_reply":"2024-03-24T02:09:00.187298Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Evaluate the model on the validation set\nval_preds = model.predict(X_test)\nval_auc = roc_auc_score(y_test, val_preds)\nprint(f'AUC: {val_auc}')","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:09:00.189729Z","iopub.execute_input":"2024-03-24T02:09:00.190018Z","iopub.status.idle":"2024-03-24T02:09:02.110118Z","shell.execute_reply.started":"2024-03-24T02:09:00.189993Z","shell.execute_reply":"2024-03-24T02:09:02.109160Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"I used rmsprop optimizer and binary cross entropy for loss. Total number of epochs is 30.","metadata":{}},{"cell_type":"markdown","source":"## XGBoost","metadata":{}},{"cell_type":"code","source":"space = {\n    'max_depth': hp.choice('max_depth', np.arange(3, 16, 1, dtype=int)),\n    'min_child_weight': hp.quniform('min_child_weight', 1, 10, 1),\n    'subsample': hp.uniform('subsample', 0.5, 1),\n    'colsample_bytree': hp.uniform('colsample_bytree', 0.5, 1),\n    'n_estimators': hp.choice('n_estimators', np.arange(100, 1000, 100, dtype=int)),\n    'learning_rate': hp.loguniform('learning_rate', np.log(0.01), np.log(0.2)),\n    'scale_pos_weight': hp.choice('scale_pos_weight', np.arange(1, 10, dtype=int)), \n}\n\n\n\ndef objective(params):\n    clf = XGBClassifier(\n        **params,\n        eval_metric='auc',\n        use_label_encoder=False,\n        silent = 1,\n        objective='binary:logistic',\n        tree_method='approx'\n    )\n    clf.fit(X_train, y_train, eval_set=[(X_test, y_test)], early_stopping_rounds=10, verbose=False)\n    preds = clf.predict_proba(X_test)[:, 1]\n    auc = roc_auc_score(y_test, preds)\n    return {'loss': -auc, 'status': STATUS_OK}\n\n\n\ntrials = Trials()\nbest = fmin(\n    fn=objective,\n    space=space,\n    algo=tpe.suggest,\n    max_evals=50, \n    trials=trials\n)","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:09:02.111546Z","iopub.execute_input":"2024-03-24T02:09:02.112013Z","iopub.status.idle":"2024-03-24T02:32:56.789615Z","shell.execute_reply.started":"2024-03-24T02:09:02.111967Z","shell.execute_reply":"2024-03-24T02:32:56.788703Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"best_params = space_eval(space, best)\nprint(\"Best parameters:\", best_params)","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:32:56.791218Z","iopub.execute_input":"2024-03-24T02:32:56.791865Z","iopub.status.idle":"2024-03-24T02:32:56.798344Z","shell.execute_reply.started":"2024-03-24T02:32:56.791831Z","shell.execute_reply":"2024-03-24T02:32:56.797667Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Initialize the model with the best hyperparameters\nmodel = XGBClassifier(**best_params)\n\n# Fit the model\nmodel.fit(X_train, y_train)\n# Predict probabilities\npred_probs = model.predict_proba(X_test)[:, 1]\n\n# Calculate AUC\nauc_score = roc_auc_score(y_test, pred_probs)\nprint(\"AUC Score:\", auc_score)","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:32:56.799671Z","iopub.execute_input":"2024-03-24T02:32:56.800270Z","iopub.status.idle":"2024-03-24T02:33:04.968265Z","shell.execute_reply.started":"2024-03-24T02:32:56.800238Z","shell.execute_reply":"2024-03-24T02:33:04.967307Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## RandomForest","metadata":{}},{"cell_type":"code","source":"from sklearn.ensemble import RandomForestClassifier\nfrom imblearn.over_sampling import SMOTE\nfrom sklearn.utils import resample","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:33:04.969435Z","iopub.execute_input":"2024-03-24T02:33:04.969757Z","iopub.status.idle":"2024-03-24T02:33:04.974199Z","shell.execute_reply.started":"2024-03-24T02:33:04.969729Z","shell.execute_reply":"2024-03-24T02:33:04.973130Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def balance_data(features, labels, sampling_strategy=0.5, random_state=42):\n    # Positive oversampling (SMOTE)\n    smote = SMOTE(sampling_strategy=sampling_strategy, random_state=random_state)\n    features_resampled, labels_resampled = smote.fit_resample(features, labels)\n\n    # Negative downsampling (reduce negative samples to match positive samples)\n    negative_samples = features_resampled[labels_resampled == 0]\n    positive_samples = features_resampled[labels_resampled == 1]\n\n    # Perform negative downsampling, e.g., reduce negative samples to match twice the number of positive samples\n    negative_samples_downsampled = resample(negative_samples, n_samples=len(positive_samples)*2, random_state=random_state)\n\n    # Combine positive and downsampled negative samples\n    balanced_features = pd.concat([positive_samples, negative_samples_downsampled], ignore_index=True)\n    balanced_labels = pd.Series([1] * len(positive_samples) + [0] * len(negative_samples_downsampled))\n\n    return balanced_features, balanced_labels\n\nX_balanced, y_balanced = balance_data(X_train, y_train, sampling_strategy=0.5, random_state=42)","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:33:04.975468Z","iopub.execute_input":"2024-03-24T02:33:04.975777Z","iopub.status.idle":"2024-03-24T02:33:06.065015Z","shell.execute_reply.started":"2024-03-24T02:33:04.975746Z","shell.execute_reply":"2024-03-24T02:33:06.064023Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"space_rf = {\n    'n_estimators': hp.choice('n_estimators', range(100, 1001, 100)),\n    'max_depth': hp.choice('max_depth', range(5, 31)),\n    'min_samples_split': hp.uniform('min_samples_split', 0.1, 1.0),\n    'min_samples_leaf': hp.uniform('min_samples_leaf', 0.1, 0.5),\n    'max_features': hp.choice('max_features', ['auto', 'sqrt', 'log2', None]),\n}\n\n\ndef objective_rf(params):\n    clf = RandomForestClassifier(**params, random_state=42)\n    score = -cross_val_score(clf, X_balanced, y_balanced, scoring='roc_auc', cv=5).mean()\n    return {'loss': score, 'status': STATUS_OK}\n\ntrials_rf = Trials()\nbest_rf = fmin(fn=objective_rf, space=space_rf, algo=tpe.suggest, max_evals=50, trials=trials_rf)\n\n\nbest_params_rf = space_eval(space_rf, best_rf)\nprint(\"Best parameters:\", best_params_rf)","metadata":{"execution":{"iopub.status.busy":"2024-03-23T10:38:46.689181Z","iopub.execute_input":"2024-03-23T10:38:46.690039Z","iopub.status.idle":"2024-03-23T11:44:32.091227Z","shell.execute_reply.started":"2024-03-23T10:38:46.690005Z","shell.execute_reply":"2024-03-23T11:44:32.090173Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"best_params_rf = {'max_depth': 11, 'max_features': None, 'min_samples_leaf': 0.12750973274384514, 'min_samples_split': 0.19805218513342315, 'n_estimators': 600}","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:33:06.066546Z","iopub.execute_input":"2024-03-24T02:33:06.066845Z","iopub.status.idle":"2024-03-24T02:33:06.071277Z","shell.execute_reply.started":"2024-03-24T02:33:06.066820Z","shell.execute_reply":"2024-03-24T02:33:06.070274Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"best_rf_model = RandomForestClassifier(**best_params_rf, random_state=42)\nbest_rf_model.fit(X_balanced, y_balanced)\n# Predict the probabilities of the positive class\ny_pred_proba = best_rf_model.predict_proba(X_test)[:, 1]\n\n# Calculate the AUC score\nauc_score = roc_auc_score(y_test, y_pred_proba)\nprint(f\"AUC Score: {auc_score:.4f}\")","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:33:06.072346Z","iopub.execute_input":"2024-03-24T02:33:06.072600Z","iopub.status.idle":"2024-03-24T02:34:48.000303Z","shell.execute_reply.started":"2024-03-24T02:33:06.072577Z","shell.execute_reply":"2024-03-24T02:34:47.999296Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## CatBoost","metadata":{}},{"cell_type":"code","source":"from catboost import CatBoostClassifier","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:34:48.001662Z","iopub.execute_input":"2024-03-24T02:34:48.001982Z","iopub.status.idle":"2024-03-24T02:34:48.006379Z","shell.execute_reply.started":"2024-03-24T02:34:48.001955Z","shell.execute_reply":"2024-03-24T02:34:48.005514Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Initialize the CatBoostClassifier\nmodel = CatBoostClassifier(\n    iterations=1000, \n    learning_rate=0.1,\n    depth=6,\n    eval_metric='AUC',\n    random_seed=42,\n    verbose=50,  \n    use_best_model=True\n)\n\n# Fit the model\nmodel.fit(\n    X_balanced, y_balanced,\n    eval_set=(X_test, y_test),\n    early_stopping_rounds=50\n)","metadata":{"execution":{"iopub.status.busy":"2024-03-23T14:20:39.177854Z","iopub.execute_input":"2024-03-23T14:20:39.178123Z","iopub.status.idle":"2024-03-23T14:20:59.348685Z","shell.execute_reply.started":"2024-03-23T14:20:39.178090Z","shell.execute_reply":"2024-03-23T14:20:59.347774Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Make predictions\ny_pred = model.predict(X_test)\n\n# Predict probabilities and calculate AUC\ny_pred_proba = model.predict_proba(X_test)[:, 1]\nauc = roc_auc_score(y_test, y_pred_proba)\nprint(f\"AUC Score: {auc:.4f}\")","metadata":{"execution":{"iopub.status.busy":"2024-03-23T12:15:29.091670Z","iopub.execute_input":"2024-03-23T12:15:29.092374Z","iopub.status.idle":"2024-03-23T12:15:29.156991Z","shell.execute_reply.started":"2024-03-23T12:15:29.092338Z","shell.execute_reply":"2024-03-23T12:15:29.156180Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Define the search space for hyperparameters\nspace = {\n    'learning_rate': hp.uniform('learning_rate', 0.01, 0.2),\n    'depth': hp.choice('depth', range(4, 10)),\n    'l2_leaf_reg': hp.uniform('l2_leaf_reg', 1, 10),\n    'iterations': hp.choice('iterations', range(100, 1000, 100)),\n    'border_count': hp.choice('border_count', range(32, 255)),\n    'loss_function': hp.choice('loss_function', ['Logloss', 'CrossEntropy']),\n}\n\ndef objective(params):\n    # Ensure parameters that should be integers are integers\n    params['depth'] = int(params['depth'])\n    params['iterations'] = int(params['iterations'])\n    params['border_count'] = int(params['border_count'])\n    \n    clf = CatBoostClassifier(**params, verbose=0, eval_metric='AUC')\n    clf.fit(X_balanced, y_balanced)\n    \n    # Predict on validation set\n    preds = clf.predict_proba(X_test)[:, 1]\n    \n    auc = roc_auc_score(y_test, preds)\n    # Return the negative AUC (because Hyperopt minimizes the objective)\n    return {'loss': -auc, 'status': STATUS_OK}","metadata":{"execution":{"iopub.status.busy":"2024-03-23T14:47:20.267448Z","iopub.execute_input":"2024-03-23T14:47:20.268291Z","iopub.status.idle":"2024-03-23T14:47:20.278296Z","shell.execute_reply.started":"2024-03-23T14:47:20.268260Z","shell.execute_reply":"2024-03-23T14:47:20.277323Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"trials = Trials()\nbest = fmin(\n    fn=objective,\n    space=space,\n    algo=tpe.suggest,\n    max_evals=50,  \n    trials=trials\n)\n\nbest_params = space_eval(space, best)\nprint(\"Best hyperparameters:\\n\", best_params)","metadata":{"execution":{"iopub.status.busy":"2024-03-23T14:47:27.072725Z","iopub.execute_input":"2024-03-23T14:47:27.073423Z","iopub.status.idle":"2024-03-23T14:57:04.922426Z","shell.execute_reply.started":"2024-03-23T14:47:27.073392Z","shell.execute_reply":"2024-03-23T14:57:04.921490Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"best_params = {'border_count': 106, 'depth': 6, 'iterations': 700, 'l2_leaf_reg': 2.419634386311497, 'learning_rate': 0.12126859554523071, 'loss_function': 'CrossEntropy'}","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:34:48.011469Z","iopub.execute_input":"2024-03-24T02:34:48.011766Z","iopub.status.idle":"2024-03-24T02:34:48.018220Z","shell.execute_reply.started":"2024-03-24T02:34:48.011731Z","shell.execute_reply":"2024-03-24T02:34:48.017206Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Initialize and train the CatBoost classifier with the best parameters\nbest_model = CatBoostClassifier(**best_params, verbose=0, eval_metric='AUC')\nbest_model.fit(X_train, y_train)\n\n# Predict probabilities on the test set\nbest_preds_proba = best_model.predict_proba(X_test)[:, 1]\n\n# Calculate AUC score\nbest_auc = roc_auc_score(y_test, best_preds_proba)\nprint(f\"Best AUC Score: {best_auc:.4f}\")","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:34:48.019178Z","iopub.execute_input":"2024-03-24T02:34:48.019429Z","iopub.status.idle":"2024-03-24T02:35:01.415439Z","shell.execute_reply.started":"2024-03-24T02:34:48.019407Z","shell.execute_reply":"2024-03-24T02:35:01.414485Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## AdaBoost","metadata":{}},{"cell_type":"code","source":"from sklearn.ensemble import AdaBoostClassifier","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:35:01.425916Z","iopub.execute_input":"2024-03-24T02:35:01.426232Z","iopub.status.idle":"2024-03-24T02:35:01.435882Z","shell.execute_reply.started":"2024-03-24T02:35:01.426201Z","shell.execute_reply":"2024-03-24T02:35:01.434938Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Initialize the base estimator\nbase_estimator = DecisionTreeClassifier(max_depth=1)\n\n# Initialize the AdaBoostClassifier\nada_clf = AdaBoostClassifier(\n    base_estimator=base_estimator,\n    n_estimators=100,  # Number of estimators to use\n    learning_rate=1.0,  # Learning rate shrinks the contribution of each classifier\n    random_state=42\n)\n\n# Fit the model on the training data\nada_clf.fit(X_balanced, y_balanced)","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:35:01.436934Z","iopub.execute_input":"2024-03-24T02:35:01.437200Z","iopub.status.idle":"2024-03-24T02:35:18.062061Z","shell.execute_reply.started":"2024-03-24T02:35:01.437178Z","shell.execute_reply":"2024-03-24T02:35:18.061098Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Make predictions\ny_pred = ada_clf.predict(X_test)\n\n# Predict probabilities for AUC\ny_pred_proba = ada_clf.predict_proba(X_test)[:, 1]\n\n# Calculate AUC\nauc = roc_auc_score(y_test, y_pred_proba)\nprint(f\"AUC Score: {auc:.4f}\")","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:35:18.063189Z","iopub.execute_input":"2024-03-24T02:35:18.063475Z","iopub.status.idle":"2024-03-24T02:35:18.679102Z","shell.execute_reply.started":"2024-03-24T02:35:18.063450Z","shell.execute_reply":"2024-03-24T02:35:18.678124Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Because the hyper-tuning of AdaBoosting is too long, we will skip this step","metadata":{}},{"cell_type":"markdown","source":"## Logistic Regression","metadata":{}},{"cell_type":"code","source":"from sklearn.linear_model import LogisticRegression","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:35:18.680282Z","iopub.execute_input":"2024-03-24T02:35:18.680570Z","iopub.status.idle":"2024-03-24T02:35:18.684993Z","shell.execute_reply.started":"2024-03-24T02:35:18.680545Z","shell.execute_reply":"2024-03-24T02:35:18.683972Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Initialize the Logistic Regression model\nlog_reg = LogisticRegression(random_state=42)\n\n# Fit the model on the training data\nlog_reg.fit(X_balanced, y_balanced)","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:35:18.686098Z","iopub.execute_input":"2024-03-24T02:35:18.686360Z","iopub.status.idle":"2024-03-24T02:35:19.990222Z","shell.execute_reply.started":"2024-03-24T02:35:18.686337Z","shell.execute_reply":"2024-03-24T02:35:19.988906Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Make predictions\ny_pred = log_reg.predict(X_test)\n\n# Predict probabilities for the positive outcome\ny_pred_proba = log_reg.predict_proba(X_test)[:, 1]\n\n# Calculate AUC\nauc = roc_auc_score(y_test, y_pred_proba)\nprint(f\"AUC Score: {auc:.4f}\")","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:35:19.996040Z","iopub.execute_input":"2024-03-24T02:35:19.997088Z","iopub.status.idle":"2024-03-24T02:35:20.046091Z","shell.execute_reply.started":"2024-03-24T02:35:19.997052Z","shell.execute_reply":"2024-03-24T02:35:20.045165Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"### Linear SVC","metadata":{}},{"cell_type":"markdown","source":"This model takes a lot of time to fit in as well as hypertuning.","metadata":{}},{"cell_type":"code","source":"from sklearn.svm import LinearSVC\nfrom sklearn.calibration import CalibratedClassifierCV","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:35:20.047544Z","iopub.execute_input":"2024-03-24T02:35:20.048254Z","iopub.status.idle":"2024-03-24T02:35:20.056524Z","shell.execute_reply.started":"2024-03-24T02:35:20.048221Z","shell.execute_reply":"2024-03-24T02:35:20.055338Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Initialize the Linear SVC model\nsvc = LinearSVC(random_state=42, max_iter=10000)\n\n# Fit the model on the scaled training data\nsvc_model = CalibratedClassifierCV(svc) \nsvc_model.fit(X_balanced, y_balanced)","metadata":{"execution":{"iopub.status.busy":"2024-03-24T03:24:46.516558Z","iopub.execute_input":"2024-03-24T03:24:46.516923Z","iopub.status.idle":"2024-03-24T03:42:45.798745Z","shell.execute_reply.started":"2024-03-24T03:24:46.516896Z","shell.execute_reply":"2024-03-24T03:42:45.797437Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Predict probabilities for the positive outcome\ny_pred_proba = svc_model.predict_proba(X_test)[:, 1]\n\n# Calculate AUC\nauc = roc_auc_score(y_test, y_pred_proba)\nprint(f\"AUC Score: {auc:.4f}\")","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:53:23.234229Z","iopub.execute_input":"2024-03-24T02:53:23.234648Z","iopub.status.idle":"2024-03-24T02:53:23.326097Z","shell.execute_reply.started":"2024-03-24T02:53:23.234595Z","shell.execute_reply":"2024-03-24T02:53:23.324790Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## REDO LightGBM","metadata":{}},{"cell_type":"markdown","source":"I will try one more time with the light GBM model with above preprocessing steps. ","metadata":{}},{"cell_type":"code","source":"# Initialize the LightGBM classifier\nlgb_clf = lgb.LGBMClassifier(\n    num_leaves=31,          # Number of leaves in one tree\n    max_depth=-1,           # Maximum tree depth for base learners, -1 means no limit\n    learning_rate=0.1,      # Learning rate\n    n_estimators=50,       # Number of boosted trees to fit\n    objective='binary',     # Objective function\n    random_state=42         # Seed for random number generator\n)\n\n# Fit the model on the training data\nlgb_clf.fit(X_balanced, y_balanced, \n            eval_set=[(X_test, y_test)],\n            eval_metric='auc')","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:53:23.335903Z","iopub.execute_input":"2024-03-24T02:53:23.339043Z","iopub.status.idle":"2024-03-24T02:53:24.738434Z","shell.execute_reply.started":"2024-03-24T02:53:23.338980Z","shell.execute_reply":"2024-03-24T02:53:24.737439Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Make predictions\ny_pred = lgb_clf.predict(X_test)\n\n# Predict probabilities for the positive class\ny_pred_proba = lgb_clf.predict_proba(X_test)[:, 1]\n\n# Calculate AUC\nauc = roc_auc_score(y_test, y_pred_proba)\nprint(f\"AUC Score: {auc:.4f}\")","metadata":{"execution":{"iopub.status.busy":"2024-03-24T02:56:14.259462Z","iopub.execute_input":"2024-03-24T02:56:14.260280Z","iopub.status.idle":"2024-03-24T02:56:14.367091Z","shell.execute_reply.started":"2024-03-24T02:56:14.260247Z","shell.execute_reply":"2024-03-24T02:56:14.366178Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Define the search space\nspace = {\n    'num_leaves': hp.choice('num_leaves', range(20, 150)),\n    'max_depth': hp.choice('max_depth', range(-1, 20)),\n    'learning_rate': hp.uniform('learning_rate', 0.01, 0.3),\n    'n_estimators': hp.choice('n_estimators', range(10, 300)),\n    'objective': hp.choice('objective', ['binary']),\n    'boosting_type': hp.choice('boosting_type', ['gbdt', 'dart', 'goss']),\n    'subsample': hp.uniform('subsample', 0.5, 1),\n    'colsample_bytree': hp.uniform('colsample_bytree', 0.5, 1),\n    'reg_alpha': hp.uniform('reg_alpha', 0, 1),\n    'reg_lambda': hp.uniform('reg_lambda', 0, 1),\n    'verbosity': -1,\n    'metric': 'auc',\n    'is_unbalance' : True\n}\n\ndef objective(params):\n    # Ensure parameters that should be integers are integers\n    params['max_depth'] = int(params['max_depth'])\n    params['n_estimators'] = int(params['n_estimators'])\n    \n    # Initialize and train the model\n    lgb_clf = lgb.LGBMClassifier(**params, random_state=42)\n    lgb_clf.fit(X_balanced, y_balanced, \n                eval_set=[(X_test, y_test)], \n                eval_metric='auc', \n                callbacks=[\n                lgb.early_stopping(stopping_rounds=10),\n                ])\n    \n    # Predict on the validation set\n    y_pred = lgb_clf.predict_proba(X_test)[:, 1]\n    \n    # Compute AUC\n    auc = roc_auc_score(y_test, y_pred)\n    \n    # Print AUC and parameters\n    print(f\"\\nAUC: {-auc:.4f}\")\n    print(f\"Params: {params}\")\n    \n    # Since hyperopt minimizes the objective, negate the auc\n    return {'loss': -auc, 'status': STATUS_OK}\nfrom hyperopt import fmin, tpe, Trials\n\n# Trials object to store iteration results\ntrials = Trials()\n\n# Run the optimization\nbest = fmin(fn=objective,\n            space=space,\n            algo=tpe.suggest,\n            max_evals=100,  # Number of evaluations\n            trials=trials)\n\nprint(\"Best hyperparameters:\\n\", best)","metadata":{"execution":{"iopub.status.busy":"2024-03-24T03:18:15.287869Z","iopub.execute_input":"2024-03-24T03:18:15.288831Z","iopub.status.idle":"2024-03-24T03:24:08.483992Z","shell.execute_reply.started":"2024-03-24T03:18:15.288780Z","shell.execute_reply":"2024-03-24T03:24:08.483127Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Adjust best parameters if necessary (for 'is_unbalance' and 'scale_pos_weight')\nbest_params = {\n    **best,\n    'objective': 'binary',\n    'metric': 'auc',\n    'boosting_type': 'gbdt',\n    'is_unbalance' : True,\n    'metric': 'auc',\n    'verbosity': -1\n}\n\n# Initialize the LightGBM classifier with the best parameters\nlgb_clf_best = lgb.LGBMClassifier(**best_params, random_state=42)\n\n# Fit the model on the training data\nlgb_clf_best.fit(X_balanced, y_balanced,\n                 eval_set=[(X_test, y_test)],\n                 eval_metric='auc', callbacks=[\n                lgb.early_stopping(stopping_rounds=10),\n                ])\n\n# Predict probabilities for the positive outcome on the test set\ny_pred_proba_best = lgb_clf_best.predict_proba(X_test)[:, 1]\n\n# Calculate AUC\nbest_auc = roc_auc_score(y_test, y_pred_proba_best)\nprint(f\"AUC Score: {best_auc:.4f}\")","metadata":{"execution":{"iopub.status.busy":"2024-03-24T03:24:18.659282Z","iopub.execute_input":"2024-03-24T03:24:18.659964Z","iopub.status.idle":"2024-03-24T03:24:22.667191Z","shell.execute_reply.started":"2024-03-24T03:24:18.659929Z","shell.execute_reply":"2024-03-24T03:24:22.666305Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Stacking Generalization Ensampling","metadata":{}}]}