{"cells":[{"metadata":{},"cell_type":"markdown","source":"# Alaska 2 LGBM Hyperparameter Optimization","execution_count":null},{"metadata":{},"cell_type":"markdown","source":"<img src=\"https://i.postimg.cc/pXzt75rk/tenor.gif\" align=\"right\" width=\"300\" height=\"200\">\n\nEveryone using CNN transfer learning for get high in LB, but why not use classic classifications techniques like binary classification with lgbm or somthing like that. \n\nIt dont work realy well but i did it in this notebook, what u gonna do ?\n\nunfortunatly we havent enough power of calculus for that ill make som transformations on images to make them edible.\n\nLets go !\n\n<BR CLEAR=”left” />","execution_count":null},{"metadata":{},"cell_type":"markdown","source":"![Sans%20titre.png](attachment:Sans%20titre.png)","attachments":{"Sans%20titre.png":{"image/png":"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"}},"execution_count":null},{"metadata":{},"cell_type":"markdown","source":"For each image we gonna apply pooling 3 x 3 x 1 on it that make images very small comparing to the original en gardant un maximum sa nature. it destroy information about steganography but we dont have choice =/.","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"import matplotlib.pyplot as plt\nimport matplotlib.image as mpimg\nimport skimage.measure\nimport numpy as np\nimport gc\n\nimg=mpimg.imread('../input/alaska2-image-steganalysis/Cover/00001.jpg')\nimgplot = plt.imshow(img)\nplt.title('Original')\nplt.show()\n\ntest_pool = skimage.measure.block_reduce(img, (3,3,1), np.max)\n\nimgplot = plt.imshow(test_pool)\nplt.title('3*3 Pooling')\nplt.show()\n\nd1, d2, d3 = test_pool.shape\ndel test_pool\ngc.collect()\n","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"U see ? it still make sense, almost nothing change ;)\n\nafter that we sum RGB colors, we flat the result and we apply dimension reduction using PCA and we give that food to lgbm for hyperparameters tuning.\n\nin reality it gives that","execution_count":null},{"metadata":{},"cell_type":"markdown","source":"![Sans%20titre%202.png](attachment:Sans%20titre%202.png)","attachments":{"Sans%20titre%202.png":{"image/png":"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"}},"execution_count":null},{"metadata":{},"cell_type":"markdown","source":"# Libraries for fun","execution_count":null},{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"import pandas as pd\nimport tqdm\nfrom PIL import Image\nimport glob\nimport lightgbm as lgb\nfrom skopt import BayesSearchCV\nfrom sklearn.decomposition import PCA\nfrom bayes_opt import BayesianOptimization\nfrom sklearn import metrics\nimport warnings\nwarnings.filterwarnings(\"ignore\")","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Images Preprocess","execution_count":null},{"metadata":{},"cell_type":"markdown","source":"I am gonna use 3 functions :\n\nalaska_weighted_auc : Competition custom metric <a href=\"https://www.kaggle.com/anokas/weighted-auc-metric-updated\">reference</a>\n\nbayes_parameter_opt_lgb : useful for select best hyperparameters for lgb using bayesian optimisation <a href=\"https://www.kaggle.com/sz8416/simple-bayesian-optimization-for-lightgbm\">reference</a>\n\nimg_reader : Import images and all process that I quoted above except PCA, made by me =D\n","execution_count":null},{"metadata":{"_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","trusted":true},"cell_type":"code","source":"def alaska_weighted_auc(y_valid, y_true):\n    tpr_thresholds = [0.0, 0.4, 1.0]\n    weights =        [       2,   1]\n\n    fpr, tpr, thresholds = metrics.roc_curve(y_true.get_label(), y_valid, pos_label=1)\n    \n    # size of subsets\n    areas = np.array(tpr_thresholds[1:]) - np.array(tpr_thresholds[:-1])\n    \n    # The total area is normalized by the sum of weights such that the final weighted AUC is between 0 and 1.\n    normalization = np.dot(areas, weights)\n    \n    competition_metric = 0\n    for idx, weight in enumerate(weights):\n        y_min = tpr_thresholds[idx]\n        y_max = tpr_thresholds[idx + 1]\n        mask = (y_min < tpr) & (tpr < y_max)\n        if mask.sum() == 0:\n            continue\n\n        x_padding = np.linspace(fpr[mask][-1], 1, 100)\n\n        x = np.concatenate([fpr[mask], x_padding])\n        y = np.concatenate([tpr[mask], [y_max] * len(x_padding)])\n        y = y - y_min # normalize such that curve starts at y=0\n        score = metrics.auc(x, y)\n        submetric = score * weight\n        best_subscore = (y_max - y_min) * weight\n        competition_metric += submetric\n        \n    return 'alaska_weighted_auc' ,competition_metric / normalization, True\n\ndef bayes_parameter_opt_lgb(X, y, init_round=15, opt_round=25, n_folds=5, random_seed=6, output_process=False):\n    # prepare data\n    train_data = lgb.Dataset(data=X, label=y, free_raw_data=False)\n    # parameters\n    def lgb_eval(num_leaves, feature_fraction, bagging_fraction, max_depth, lambda_l1, lambda_l2, min_split_gain, min_child_weight, learning_rate, n_estimators):\n        params = {'application':'binary', 'early_stopping_round':100, 'metric':'auc', 'objective' : 'binary'}\n        params[\"num_leaves\"] = int(round(num_leaves))\n        params['feature_fraction'] = max(min(feature_fraction, 1), 0)\n        params['bagging_fraction'] = max(min(bagging_fraction, 1), 0)\n        params['max_depth'] = int(round(max_depth))\n        params['lambda_l1'] = max(lambda_l1, 0)\n        params['lambda_l2'] = max(lambda_l2, 0)\n        params['min_split_gain'] = min_split_gain\n        params['min_child_weight'] = min_child_weight\n        params['learning_rate'] = max(min(learning_rate, 1), 0.001)\n        params['n_estimators'] = int(round(n_estimators))\n        cv_result = lgb.cv(params, train_data, nfold=n_folds, seed=random_seed, stratified=True, verbose_eval =200, metrics=['auc'], feval = alaska_weighted_auc)\n        return max(cv_result['auc-mean'])\n    # range \n    lgbBO = BayesianOptimization(lgb_eval, {'num_leaves': (10, 80),\n                                            'feature_fraction': (0.1, 0.9),\n                                            'bagging_fraction': (0.6, 1),\n                                            'max_depth': (5, 20),\n                                            'lambda_l1': (0, 10),\n                                            'lambda_l2': (0, 10),\n                                            'min_split_gain': (0.001, 0.1),\n                                            'min_child_weight': (5, 50),\n                                            'learning_rate' : (0.001, 0.1),\n                                            'n_estimators' : (100, 10000)}, random_state=0)\n    \n    # optimize\n    lgbBO.maximize(init_points=init_round, n_iter=opt_round)\n    \n    # output optimization process\n    if output_process==True: lgbBO.points_to_csv(\"bayes_opt_result.csv\")\n    \n    # return best parameters\n    return lgbBO\n\ndef img_reader(nbr_images = 10, df = None, file_name = 'Cover', from_ = 0, status = 'neg') :\n    from_ = from_\n    nbr_images  = nbr_images\n    image_list = []\n    i=0\n    j=0\n    df = df\n    file_name = file_name\n    for filename in tqdm.tqdm(glob.glob('../input/alaska2-image-steganalysis/'+file_name+'/*.jpg')): \n        if j >= from_ :\n            im=mpimg.imread(filename)\n            im=skimage.measure.block_reduce(im, (3,3,1), np.max)\n            image_list.append(np.sum(im.reshape((d3, d1*d2)), axis = 0).tolist())\n            i+=1\n            if i%1000 == 0 :\n                if df is None:\n                    df = pd.DataFrame(image_list).astype('int16')\n                    del image_list\n                    gc.collect()\n                    image_list = []\n                else :\n                    df = pd.concat([df , pd.DataFrame(image_list).astype('int16')])\n                    del image_list\n                    gc.collect()\n                    image_list = []\n                    if i == nbr_images :    \n                        del image_list\n                        gc.collect()\n                        break\n        j=j+1\n        \n    if status == 'neg' :\n        df['output'] = 0\n        df['output'] = df['output'].astype('int16')\n        gc.collect()\n    else :\n        df['output'] = 1\n        df['output'] = df['output'].astype('int16')\n        gc.collect()\n        \n    return df","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"img=mpimg.imread('../input/alaska2-image-steganalysis/Cover/00001.jpg')\ntest_pool = skimage.measure.block_reduce(img, (3,3,1), np.max)\nd1, d2, d3 = test_pool.shape\ndel test_pool\ngc.collect()\n\ndf_neg = img_reader(nbr_images = 12000, df = None, file_name = 'Cover', from_ = 0, status = 'neg')\n\ndf_pos = img_reader(nbr_images = 4000, df = None, file_name = 'JMiPOD', from_ = 0, status = 'pos')\nprint('JMiPOD Done!')\ndf_pos = img_reader(nbr_images = 4000, df = df_pos, file_name = 'JUNIWARD', from_ = 4000, status = 'pos')\nprint('JUNIWARD Done!')\ndf_pos = img_reader(nbr_images = 4000, df = df_pos, file_name = 'UERD', from_ = 8000, status = 'pos')\nprint('UERD Done!')\n\ndf_test = img_reader(nbr_images = 6000, df = None, file_name = 'Test', from_ = 0, status = 'neg')\nprint('Test Done!')","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"Save result as pkl format for later.","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"df_train = pd.concat([df_pos, df_neg], ignore_index = True)\ndel df_pos, df_neg\n\ndf_train.to_pickle('df_train3*3.pkl')\ndf_test.to_pickle('df_test3*3.pkl')\n\ndel df_train, df_test\ngc.collect()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"df_train = pd.read_pickle('./df_train3*3.pkl')\ndf_test = pd.read_pickle('./df_test3*3.pkl')","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"Apply PCA with 500 components","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"pca1 = PCA(n_components=500)\ndf_train_pca = pca1.fit_transform(df_train.loc[:, df_train.columns != 'output'].values)\n\npca2 = PCA(n_components=500)\ndf_test_pca = pca2.fit_transform(df_test.loc[:, df_test.columns != 'output'].values)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"X = df_train_pca\ny = df_train['output']\ndel df_train_pca ,\ngc.collect()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Hyperparameter Optimization","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"opt_params = bayes_parameter_opt_lgb(X, y, init_round=15, opt_round=30, n_folds=5, random_seed=6)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"print('Best Params :')\n\nprint(opt_params.max['params'])","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Train LGBM","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"params = opt_params.max['params']\nparams['num_leaves'] = int(params['num_leaves'])\nparams['max_depth'] = int(params['max_depth'])\nparams['n_estimators'] = int(params['n_estimators'])\n\nd_train = lgb.Dataset(data=X, label=y, free_raw_data=False)\n\nclf = lgb.train(params, train_set = d_train,  feval = alaska_weighted_auc)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"lgb.plot_importance(clf, max_num_features = 10)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"lgb.create_tree_digraph(clf)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Predict and Submit","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"y_pred=clf.predict(df_test_pca)\nsub = pd.read_csv('../input/alaska2-image-steganalysis/sample_submission.csv')\nsub['Label'] = y_pred\nsub.to_csv('submission.csv', index=False)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"Thanks for reading my notebook, if you have any suggestion i will be happy to receive it.\n\nThis solution don't work well andit will not propel you to the top of the LB so dont UpVote.","execution_count":null}],"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat":4,"nbformat_minor":4}