04_c__FER_SVM_PCA.ipynb

{"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_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"# Setup","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19"}},{"cell_type":"code","source":"# setup to see the execution time in each cell\n\n!pip install ipython-autotime\n!pip install directory_structure\n# !pip install wandb\n%load_ext autotime","metadata":{"execution":{"iopub.status.busy":"2022-05-09T05:57:13.241002Z","iopub.execute_input":"2022-05-09T05:57:13.241461Z","iopub.status.idle":"2022-05-09T05:57:28.787744Z","shell.execute_reply.started":"2022-05-09T05:57:13.241322Z","shell.execute_reply":"2022-05-09T05:57:28.786902Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import pandas as pd\nimport os\nimport glob\nimport PIL\nfrom PIL import Image\nimport numpy as np\nimport cupy as cp\nimport matplotlib.pyplot as plt","metadata":{"execution":{"iopub.status.busy":"2022-05-09T05:57:28.790846Z","iopub.execute_input":"2022-05-09T05:57:28.791652Z","iopub.status.idle":"2022-05-09T05:57:31.076186Z","shell.execute_reply.started":"2022-05-09T05:57:28.791613Z","shell.execute_reply":"2022-05-09T05:57:31.075398Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"DRIVER_ROOT_DIR = \"/kaggle/input/\" # HC Directory\n# DRIVER_ROOT_DIR = \"/content/drive/MyDrive/DTSC 870/Code\" #MT Directory\n\n# no augmentation\n# DATASET_02_TRAIN = DRIVER_ROOT_DIR + \"fer2013/02_FER/train\"\n# DATASET_02_TEST = DRIVER_ROOT_DIR + \"fer2013/02_FER/test\"\n\n# augmentation v.1\n# DATASET_02_TRAIN = DRIVER_ROOT_DIR + \"fer2013-aug/Aug_train\"\n# DATASET_02_TEST = DRIVER_ROOT_DIR + \"fer2013-aug-test/Aug_test\"\n\n# augmentation v.2\nDATASET_02_TRAIN = DRIVER_ROOT_DIR + \"fer2013-aug-train-2/Aug_train_2\"\n# DATASET_02_TEST = DRIVER_ROOT_DIR + \"fer2013-aug-test-2/Aug_test_2\"\nDATASET_02_TEST = DRIVER_ROOT_DIR + \"d/datasets/huihenrychen/fer2013-aug-test-2/Aug_test_2\"","metadata":{"execution":{"iopub.status.busy":"2022-05-09T05:57:31.077446Z","iopub.execute_input":"2022-05-09T05:57:31.077734Z","iopub.status.idle":"2022-05-09T05:57:31.08365Z","shell.execute_reply.started":"2022-05-09T05:57:31.077678Z","shell.execute_reply":"2022-05-09T05:57:31.082952Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Utility Functions","metadata":{}},{"cell_type":"code","source":"def get_data_df(dir):\n    # modified code from: https://www.kaggle.com/namgalielei/simple-load-images-and-count-number-of-each-class\n\n    train_df = pd.DataFrame()\n\n    trainset = glob.glob(dir)\n\n    train_df['file'] = [img.split(\"/\")[-1] for img in trainset]\n    train_df['class'] = [img.split(\"/\")[-2] for img in trainset]\n\n    return train_df","metadata":{"execution":{"iopub.status.busy":"2022-05-09T05:57:31.084791Z","iopub.execute_input":"2022-05-09T05:57:31.085404Z","iopub.status.idle":"2022-05-09T05:57:31.094136Z","shell.execute_reply.started":"2022-05-09T05:57:31.085365Z","shell.execute_reply":"2022-05-09T05:57:31.093276Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def generate_set(df, dir, classes_):\n\n    # new_df = pd.DataFrame()\n    pixels = []\n    class_ = []\n\n    # trainset = glob.glob(dir)\n    for i in range(len(df.index)):\n        # get the absolute img path\n        # e.g., Brain_tumor_images/<train or test>/<class label>/<file name>\n        path = dir + \"/\" + df.iloc[i][\"class\"] + \"/\" +df.iloc[i][\"file\"]\n        # print(img)\n        img = Image.open(path)\n        # print(\"Img: {} \\tClass: {}\".format(np.array(img).flatten(), df.iloc[i][\"class\"]))\n        pixels.append(cp.asnumpy(cp.array(img)).flatten())\n        # pixels.append(np.array(img))\n\n        # y_true encoding here\n        class_.append(classes_.index(df.iloc[i][\"class\"]))\n\n        # end loop here\n\n    # return train_df\n    return pixels, class_","metadata":{"execution":{"iopub.status.busy":"2022-05-09T05:57:31.096862Z","iopub.execute_input":"2022-05-09T05:57:31.097199Z","iopub.status.idle":"2022-05-09T05:57:31.105002Z","shell.execute_reply.started":"2022-05-09T05:57:31.097164Z","shell.execute_reply":"2022-05-09T05:57:31.104037Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Generate the train and test sets","metadata":{}},{"cell_type":"code","source":"fer_df_train = get_data_df(DATASET_02_TRAIN+\"/*/*.jpg\")\nfer_df_test = get_data_df(DATASET_02_TEST+\"/*/*.jpg\")","metadata":{"execution":{"iopub.status.busy":"2022-05-09T05:57:31.106567Z","iopub.execute_input":"2022-05-09T05:57:31.107288Z","iopub.status.idle":"2022-05-09T05:57:33.16618Z","shell.execute_reply.started":"2022-05-09T05:57:31.107244Z","shell.execute_reply":"2022-05-09T05:57:33.165389Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"classes = fer_df_train[\"class\"].unique().tolist()\nclasses\n# classes.index(\"surprise\")","metadata":{"execution":{"iopub.status.busy":"2022-05-09T05:57:33.167411Z","iopub.execute_input":"2022-05-09T05:57:33.168862Z","iopub.status.idle":"2022-05-09T05:57:33.181165Z","shell.execute_reply.started":"2022-05-09T05:57:33.168819Z","shell.execute_reply":"2022-05-09T05:57:33.180361Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"x_train, y_train = generate_set(fer_df_train, DATASET_02_TRAIN, classes)\nx_test, y_test = generate_set(fer_df_test, DATASET_02_TEST, classes)","metadata":{"execution":{"iopub.status.busy":"2022-05-09T05:57:33.182829Z","iopub.execute_input":"2022-05-09T05:57:33.183321Z","iopub.status.idle":"2022-05-09T06:01:37.65522Z","shell.execute_reply.started":"2022-05-09T05:57:33.183284Z","shell.execute_reply":"2022-05-09T06:01:37.6544Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"### EDA","metadata":{}},{"cell_type":"code","source":"# fer_df_test.sample(10)","metadata":{"execution":{"iopub.status.busy":"2022-05-09T06:01:37.65654Z","iopub.execute_input":"2022-05-09T06:01:37.656831Z","iopub.status.idle":"2022-05-09T06:01:37.663403Z","shell.execute_reply.started":"2022-05-09T06:01:37.656794Z","shell.execute_reply":"2022-05-09T06:01:37.662379Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# fer_df_train.shape","metadata":{"execution":{"iopub.status.busy":"2022-05-09T06:01:37.665141Z","iopub.execute_input":"2022-05-09T06:01:37.665728Z","iopub.status.idle":"2022-05-09T06:01:37.673576Z","shell.execute_reply.started":"2022-05-09T06:01:37.665673Z","shell.execute_reply":"2022-05-09T06:01:37.672838Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"x_train_df = pd.DataFrame()\nx_train_df['class'] = y_train\n\n# x_train_df","metadata":{"execution":{"iopub.status.busy":"2022-05-09T06:01:37.675357Z","iopub.execute_input":"2022-05-09T06:01:37.675981Z","iopub.status.idle":"2022-05-09T06:01:37.717648Z","shell.execute_reply.started":"2022-05-09T06:01:37.675942Z","shell.execute_reply":"2022-05-09T06:01:37.7169Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"x_train_df[\"class\"].hist()","metadata":{"execution":{"iopub.status.busy":"2022-05-09T06:01:37.718722Z","iopub.execute_input":"2022-05-09T06:01:37.719458Z","iopub.status.idle":"2022-05-09T06:01:37.957803Z","shell.execute_reply.started":"2022-05-09T06:01:37.71942Z","shell.execute_reply":"2022-05-09T06:01:37.956685Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# print(np.array(x_train).shape)\n# print(np.array(x_test).shape)","metadata":{"execution":{"iopub.status.busy":"2022-05-09T06:01:37.95916Z","iopub.execute_input":"2022-05-09T06:01:37.959403Z","iopub.status.idle":"2022-05-09T06:01:37.964027Z","shell.execute_reply.started":"2022-05-09T06:01:37.959368Z","shell.execute_reply":"2022-05-09T06:01:37.963223Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Class weights","metadata":{}},{"cell_type":"code","source":"print(\"Classes: {}\".format(classes))\nprint(\"X train shape: {}\".format(np.array(x_train).shape))\nprint(\"X test shape: {}\".format(np.array(x_test).shape))\n\n# total samples\nN = np.array(x_train).shape[0] + np.array(x_test).shape[0]\nprint(\"Total sample size: {}\".format(N))\n\n\n# total class sample count\ntrain_set_count = np.bincount(np.array(y_train))\ntest_set_count = np.bincount(np.array(y_test))\nprint(\"Train set: {}\".format(train_set_count))\nprint(\"Test set: {}\".format(test_set_count))\n# print(type(test_set_count))\n\n\n# assign weights to each classes\nc_weight = {}\nclass_len = len(classes)\nfor i in range(len(classes)):\n    c_weight[i] = N/(class_len*(train_set_count[i] + test_set_count[i]))\n    \nprint(\"The class weights: {}\".format(c_weight))","metadata":{"execution":{"iopub.status.busy":"2022-05-09T06:01:37.968181Z","iopub.execute_input":"2022-05-09T06:01:37.968721Z","iopub.status.idle":"2022-05-09T06:01:38.355593Z","shell.execute_reply.started":"2022-05-09T06:01:37.968671Z","shell.execute_reply":"2022-05-09T06:01:38.354727Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Feature Engineering","metadata":{}},{"cell_type":"code","source":"# from sklearn.decomposition import PCA\nfrom cuml.decomposition import PCA\nimport matplotlib.pyplot as plt\nimport plotly.express as px\nfrom sklearn.preprocessing import StandardScaler\nimport cv2\nfrom skimage.feature import local_binary_pattern\nfrom plotly.subplots import make_subplots\nimport plotly.graph_objects as go\nfrom matplotlib.pyplot import figure","metadata":{"execution":{"iopub.status.busy":"2022-05-09T06:01:38.356855Z","iopub.execute_input":"2022-05-09T06:01:38.357656Z","iopub.status.idle":"2022-05-09T06:01:41.42558Z","shell.execute_reply.started":"2022-05-09T06:01:38.357616Z","shell.execute_reply":"2022-05-09T06:01:41.424774Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## PCA","metadata":{}},{"cell_type":"code","source":"pca = PCA()\nx_train_pca = pca.fit_transform(np.array(x_train, dtype=np.float32))\nx_test_pca = pca.transform(np.array(x_test, dtype=np.float32))","metadata":{"execution":{"iopub.status.busy":"2022-05-09T06:01:41.427064Z","iopub.execute_input":"2022-05-09T06:01:41.427563Z","iopub.status.idle":"2022-05-09T06:01:53.587371Z","shell.execute_reply.started":"2022-05-09T06:01:41.427524Z","shell.execute_reply":"2022-05-09T06:01:53.586418Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"exp_var_cumul = np.cumsum(pca.explained_variance_ratio_)\nexp_var_cumul_round = np.round_(exp_var_cumul, decimals = 4)\n\nnum_comp = range(1, exp_var_cumul_round.shape[0] + 1)\n\nprint(exp_var_cumul_round)","metadata":{"execution":{"iopub.status.busy":"2022-05-09T06:01:53.588819Z","iopub.execute_input":"2022-05-09T06:01:53.589088Z","iopub.status.idle":"2022-05-09T06:01:53.597201Z","shell.execute_reply.started":"2022-05-09T06:01:53.589052Z","shell.execute_reply":"2022-05-09T06:01:53.596391Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"### PCA Visualization","metadata":{}},{"cell_type":"code","source":"labels = {\n    str(i): f\"PC {i+1} ({var:.1f}%)\"\n    for i, var in enumerate(pca.explained_variance_ratio_ * 100)\n}\n\nfig = px.scatter_matrix(\n    x_train_pca,\n    # x_train_df,\n    labels=labels,\n    # dimensions=(0, 1),\n    dimensions=range(4),\n    # color=x_train_df['class'],\n)\nfig.update_traces(diagonal_visible=False)\nfig.show()","metadata":{"scrolled":true,"execution":{"iopub.status.busy":"2022-05-09T06:01:53.598793Z","iopub.execute_input":"2022-05-09T06:01:53.599351Z","iopub.status.idle":"2022-05-09T06:01:54.695089Z","shell.execute_reply.started":"2022-05-09T06:01:53.599256Z","shell.execute_reply":"2022-05-09T06:01:54.692791Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# get a list of # of components\nnum_comp = range(1, exp_var_cumul_round.shape[0] + 1)\n\nper_var = np.round(pca.explained_variance_ratio_* 100, decimals=1)\n\nfig = px.line(\n    x=num_comp,\n    y=exp_var_cumul_round,\n    labels={\"x\": \"# Components\", \"y\": \"Cumulative Explained Variance\"},\n    title = \"# of components V.S. variance\",\n    # markers=True\n)\n\nfig.show()","metadata":{"execution":{"iopub.status.busy":"2022-05-09T06:01:54.69604Z","iopub.execute_input":"2022-05-09T06:01:54.696278Z","iopub.status.idle":"2022-05-09T06:01:54.821296Z","shell.execute_reply.started":"2022-05-09T06:01:54.696247Z","shell.execute_reply":"2022-05-09T06:01:54.820607Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# selected_exp_var = [0.7, 0.8, 0.9, 0.95, 0.97, 0.99, 1.0]\n# selected_num_comp_aug = [14, 32, 108, 269, 446, 940, 2265]","metadata":{"execution":{"iopub.status.busy":"2022-05-09T06:01:54.822629Z","iopub.execute_input":"2022-05-09T06:01:54.82306Z","iopub.status.idle":"2022-05-09T06:01:54.827941Z","shell.execute_reply.started":"2022-05-09T06:01:54.823021Z","shell.execute_reply":"2022-05-09T06:01:54.827031Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"per_var = np.round(pca.explained_variance_ratio_* 100, decimals=1)\n# labels = ['PC' + str(x) for x in range(1, len(per_var)+1)]\n\nfig = px.bar(\n    x=range(1,len(per_var)+1), \n    y=per_var, \n    title='Scree Plot of # of components and % of explained variance',\n    labels={\"x\": \"# Components\", \"y\": \"% of Explained Variance\"},\n)\nfig.show()","metadata":{"execution":{"iopub.status.busy":"2022-05-09T06:01:54.829495Z","iopub.execute_input":"2022-05-09T06:01:54.830102Z","iopub.status.idle":"2022-05-09T06:01:54.917519Z","shell.execute_reply.started":"2022-05-09T06:01:54.830064Z","shell.execute_reply":"2022-05-09T06:01:54.916688Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Model","metadata":{}},{"cell_type":"markdown","source":"## Setup","metadata":{}},{"cell_type":"code","source":"from sklearn.metrics import balanced_accuracy_score, accuracy_score\nfrom numpy import mean\n\nfrom cuml.svm import SVC\n# from sklearn.svm import SVC","metadata":{"execution":{"iopub.status.busy":"2022-05-09T06:01:54.918979Z","iopub.execute_input":"2022-05-09T06:01:54.919221Z","iopub.status.idle":"2022-05-09T06:01:54.924823Z","shell.execute_reply.started":"2022-05-09T06:01:54.91919Z","shell.execute_reply":"2022-05-09T06:01:54.923928Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def generate_PCA_set(comp, x_train, x_test):\n    pca = PCA(n_components=comp)\n    __x_train__ = pca.fit_transform(x_train)\n    __x_test__ = pca.transform(x_test)\n    return __x_train__, __x_test__","metadata":{"execution":{"iopub.status.busy":"2022-05-09T06:01:54.926005Z","iopub.execute_input":"2022-05-09T06:01:54.926527Z","iopub.status.idle":"2022-05-09T06:01:54.934414Z","shell.execute_reply.started":"2022-05-09T06:01:54.926487Z","shell.execute_reply":"2022-05-09T06:01:54.933522Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def get_model():\n    return SVC(random_state=1, multiclass_strategy=\"ovr\")\n\ndef get_model2():\n    return SVC(random_state=1, multiclass_strategy=\"ovr\", class_weight=c_weight)","metadata":{"execution":{"iopub.status.busy":"2022-05-09T06:01:54.93629Z","iopub.execute_input":"2022-05-09T06:01:54.936882Z","iopub.status.idle":"2022-05-09T06:01:54.943381Z","shell.execute_reply.started":"2022-05-09T06:01:54.936846Z","shell.execute_reply":"2022-05-09T06:01:54.942481Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def feature_scale(_x_train_, _x_test_):\n    sc = StandardScaler()\n    \n    train_sc = sc.fit_transform(_x_train_)\n    test_sc = sc.transform(_x_test_)\n    \n    return train_sc, test_sc\n\n\ndef feature_scale2(_x_train_):\n    sc = StandardScaler()\n    \n    train_sc2 = sc.fit_transform(_x_train_)\n    \n    return train_sc2","metadata":{"execution":{"iopub.status.busy":"2022-05-09T06:01:54.944972Z","iopub.execute_input":"2022-05-09T06:01:54.945636Z","iopub.status.idle":"2022-05-09T06:01:54.954019Z","shell.execute_reply.started":"2022-05-09T06:01:54.945597Z","shell.execute_reply":"2022-05-09T06:01:54.953227Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Train","metadata":{}},{"cell_type":"code","source":"selected_exp_var = [0.7, 0.8, 0.9, 0.95, 0.97, 0.99, 1.0]\n\n# none img augmentation\nselected_num_comp = [13, 32, 104, 256, 425, 904, 2304]\n\n# img augmentation\nselected_num_comp_aug = [14, 32, 108, 269, 446, 940, 2265]","metadata":{"execution":{"iopub.status.busy":"2022-05-09T06:01:54.955586Z","iopub.execute_input":"2022-05-09T06:01:54.956219Z","iopub.status.idle":"2022-05-09T06:01:54.96289Z","shell.execute_reply.started":"2022-05-09T06:01:54.95618Z","shell.execute_reply":"2022-05-09T06:01:54.961885Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"### Training with No Class weights","metadata":{}},{"cell_type":"code","source":"svm_pca_experiment = list()\n\nfor i in range(len(selected_exp_var)):\n    \n#     temp_train, temp_test = generate_PCA_set(selected_num_comp[i], x_train, x_test)\n    temp_train, temp_test = generate_PCA_set(selected_num_comp_aug[i], np.array(x_train, dtype=np.float32), np.array(x_test, dtype=np.float32))\n    \n    svm_pca_experiment.append({\n        \"num_compon\": selected_num_comp[i],\n        \"variance\": selected_exp_var[i],\n        \"model\": get_model(),\n        \"x_train\": temp_train,\n        \"x_test\": temp_test\n    })","metadata":{"execution":{"iopub.status.busy":"2022-05-09T06:01:54.963923Z","iopub.execute_input":"2022-05-09T06:01:54.965833Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"### Training with Class weights","metadata":{}},{"cell_type":"code","source":"# svm_pca_experiment2 = svm_pca_experiment\n\n# for i in range(len(selected_exp_var)):\n    \n#     svm_pca_experiment2[i][\"model\"] = get_model2()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def svm_pca_training(arr_, y_train, y_test):\n    uw_acc_test_result = list()\n    w_acc_test_result = list()\n    y_pred = list()\n    \n    for i in range(len(arr_)):\n        # scaling\n        x_train_sc, x_test_sc = feature_scale(arr_[i][\"x_train\"], arr_[i][\"x_test\"])\n\n        x_train_sc, x_test_sc = np.array(x_train_sc), np.array(x_test_sc)\n        y_train = np.array(y_train, dtype=np.float32)\n\n        # train the model\n        # model = SVC(C=0.1, gamma=0.01, kernel='rbf', random_state=1, multiclass_strategy=\"ovr\")\n        model = arr_[i][\"model\"]\n        \n        # model = SVC(decision_function_shape=\"ovr\", kernel=\"rbf\", random_state=13)\n        model.fit(x_train_sc, y_train)\n\n        # predict the train set\n        y_trainHat = model.predict(x_train_sc)\n\n        # predict the train set\n        y_testHat = model.predict(x_test_sc)\n        \n        y_pred.append(y_testHat)\n\n        # compute the unweighted test accuracy\n        uw_acc = balanced_accuracy_score(y_test, y_testHat)\n        uw_acc_test_result.append(uw_acc)\n        print(\"The unweighted accuracy: {}\".format(uw_acc))\n\n        # compute the weighted test accuracy\n        w_acc = accuracy_score(y_test, y_testHat)\n        w_acc_test_result.append(w_acc)\n        print(\"The weighted accuracy: {}\".format(w_acc))\n        print(\"\\n\")\n        \n    \n    return uw_acc_test_result, w_acc_test_result, y_pred\n\n\n\ndef svm_pca_training2(arr_, y_train, y_test):\n    uw_acc_test_result = list()\n    w_acc_test_result = list()\n    y_pred = list()\n\n    # scaling\n    x_train_sc, x_test_sc = feature_scale(arr_[\"x_train\"], arr_[\"x_test\"])\n\n    x_train_sc, x_test_sc = np.array(x_train_sc), np.array(x_test_sc)\n    y_train = np.array(y_train, dtype=np.float32)\n\n    # train the model\n    # model = SVC(C=0.1, gamma=0.01, kernel='rbf', random_state=1, multiclass_strategy=\"ovr\")\n    model = arr_[\"model\"]\n\n    # model = SVC(decision_function_shape=\"ovr\", kernel=\"rbf\", random_state=13)\n    model.fit(x_train_sc, y_train)\n\n    # predict the train set\n    y_trainHat = model.predict(x_train_sc)\n\n    # predict the train set\n    y_testHat = model.predict(x_test_sc)\n\n    y_pred.append(y_testHat)\n\n    # compute the unweighted test accuracy\n    uw_acc = balanced_accuracy_score(y_test, y_testHat)\n    uw_acc_test_result.append(uw_acc)\n    print(\"The unweighted accuracy: {}\".format(uw_acc))\n\n    # compute the weighted test accuracy\n    w_acc = accuracy_score(y_test, y_testHat)\n    w_acc_test_result.append(w_acc)\n    print(\"The weighted accuracy: {}\".format(w_acc))\n    print(\"\\n\")\n        \n    \n    return uw_acc_test_result, w_acc_test_result, y_pred","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# uw_acc_result, w_acc_result, y_pred = svm_pca_training2(svm_pca_experiment[2], y_train, y_test)\nuw_acc_result, w_acc_result, y_pred = svm_pca_training(svm_pca_experiment, y_train, y_test)\n\n# uw_acc_result, w_acc_result, y_pred = svm_pca_training(svm_pca_experiment2, y_train, y_test)","metadata":{"execution":{"iopub.status.busy":"2022-05-09T05:51:25.691644Z","iopub.status.idle":"2022-05-09T05:51:25.692531Z","shell.execute_reply.started":"2022-05-09T05:51:25.692294Z","shell.execute_reply":"2022-05-09T05:51:25.692321Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Model Train","metadata":{}},{"cell_type":"code","source":"from sklearn.multiclass import OneVsRestClassifier","metadata":{"execution":{"iopub.status.busy":"2022-05-09T05:51:25.69362Z","iopub.status.idle":"2022-05-09T05:51:25.694493Z","shell.execute_reply.started":"2022-05-09T05:51:25.694248Z","shell.execute_reply":"2022-05-09T05:51:25.694283Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"np.unique(c_weight)","metadata":{"execution":{"iopub.status.busy":"2022-05-09T05:51:25.695574Z","iopub.status.idle":"2022-05-09T05:51:25.696477Z","shell.execute_reply.started":"2022-05-09T05:51:25.696228Z","shell.execute_reply":"2022-05-09T05:51:25.696254Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def train_model2(arr_, y_train, y_test):\n    y_pred2 = list()\n    \n    # scaling\n    x_train_sc, x_test_sc = feature_scale(arr_[\"x_train\"], arr_[\"x_test\"])\n\n    x_train_sc, x_test_sc = np.array(x_train_sc), np.array(x_test_sc)\n    y_train = np.array(y_train, dtype=np.float32)\n\n    # train the model\n    # model = SVC(C=0.1, gamma=0.01, kernel='rbf', random_state=1, multiclass_strategy=\"ovr\")\n    model = OneVsRestClassifier(SVC(random_state=1, multiclass_strategy=\"ovr\"))\n\n    # model = SVC(decision_function_shape=\"ovr\", kernel=\"rbf\", random_state=13)\n    model.fit(x_train_sc, y_train)\n\n    # predict the train set\n    y_trainHat = model.predict(x_train_sc)\n\n    # predict the train set\n    y_testHat = model.predict(x_test_sc)\n\n    y_pred2.append(y_testHat)\n    \n    # calculate the gamma value\n    n_features = x_train_sc.shape[1]\n    gamma = 1/(n_features * x_train_sc.var())\n    print('Gamma: %.09f' % gamma)\n\n    # compute the unweighted test accuracy\n    uw_acc = balanced_accuracy_score(y_test, y_testHat)\n    # uw_acc_test_result.append(uw_acc)\n    print(\"The unweighted accuracy: {}\".format(uw_acc))\n\n    # compute the weighted test accuracy\n    w_acc = accuracy_score(y_test, y_testHat)\n    # w_acc_test_result.append(w_acc)\n    print(\"The weighted accuracy: {}\".format(w_acc))\n    print(\"\\n\")\n    \n    return model, y_pred2\n    \n\nmodel2, y_pred2 = train_model2(svm_pca_experiment[2], y_train, y_test)","metadata":{"execution":{"iopub.status.busy":"2022-05-09T05:51:25.697636Z","iopub.status.idle":"2022-05-09T05:51:25.698524Z","shell.execute_reply.started":"2022-05-09T05:51:25.698286Z","shell.execute_reply":"2022-05-09T05:51:25.698312Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Result Analysis","metadata":{}},{"cell_type":"code","source":"# no augmentation\nuw_acc_result = [0.34913545500412424, 0.410326622837653, 0.4606871805554088, 0.4416885384206618, 0.42643337815184973, 0.39500145483379384, 0.35858548421567005]\nw_acc_result = [0.36458623572025634, 0.44831429367511844, 0.49080523822791866, 0.48021733073279466, 0.4653106714962385, 0.4395374756199498, 0.40025076623014766]\n\n# augmentation 1\n# uw_acc_result = [0.30401166995891815, 0.36459252004568043, 0.3921733369630049, 0.37487205858731293, 0.3576369966351716, 0.327007717935318, 0.2870783470228984]\n# w_acc_result = [0.34159933129005293, 0.40707718027305656, 0.43076065756478127, 0.42073000835887436, 0.40373363053775424, 0.3757314015045974, 0.3354694901086654]\n\n# augmentation 2\n# uw_acc_result = [0.401166992591892, 0.36459252004568043, 0.40217333978230513, 0.38487135815731293, 0.33552365266351716, 0.303055725934518, 0.2752783475228984]\n# w_acc_result = [0.4159253129002493, 0.40707718027305656, 0.44076985756473527, 0.40073550832551436, 0.38075363053775424, 0.349514015045974, 0.3414414901063654]","metadata":{"execution":{"iopub.status.busy":"2022-05-09T05:45:52.803498Z","iopub.execute_input":"2022-05-09T05:45:52.803856Z","iopub.status.idle":"2022-05-09T05:45:52.827609Z","shell.execute_reply.started":"2022-05-09T05:45:52.80378Z","shell.execute_reply":"2022-05-09T05:45:52.827002Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# import plotly.graph_objects as go\n\n# fig = go.Figure()\n# fig.add_trace(go.Line(\n#     x=selected_exp_var,\n#     y=uw_acc_result,\n#     mode='lines',\n#     name='Unweighted Test Accuracy',\n#     marker=dict(\n#         color='red',\n#         size=10\n#     ))\n# )\n\n# fig.add_trace(go.Line(\n#     x=selected_exp_var,\n#     y=w_acc_result,\n#     mode='lines',\n#     name='Weighted Test Accuracy',\n#     marker=dict(\n#         color='green',\n#         size=10\n#     ))\n# )\n\n\n# fig.update_layout(\n#     title=\"SVM accuracy of different PCA variance (no aug)\",\n#     xaxis_title=\"SVM % of Explained Variance\",\n#     yaxis_title=\"Accuracy (%)\"\n# )\n# fig.update_traces(mode='markers+lines')\n\n# fig.show()","metadata":{"execution":{"iopub.status.busy":"2022-05-04T12:48:42.168018Z","iopub.execute_input":"2022-05-04T12:48:42.168515Z","iopub.status.idle":"2022-05-04T12:48:42.180851Z","shell.execute_reply.started":"2022-05-04T12:48:42.168475Z","shell.execute_reply":"2022-05-04T12:48:42.179975Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# plt.bar(x=\"unweighted\", height=uw_acc)\n# plt.bar(x=\"weighted\", height=w_acc)\n# plt.title(\"HoG on img augmentation SVM accuracy\")\n# plt.ylabel(\"Accuracy (%)\")\n# plt.xlabel(\"SVM accuracy type\")\n# plt.show()","metadata":{"execution":{"iopub.status.busy":"2022-05-04T12:48:42.182164Z","iopub.execute_input":"2022-05-04T12:48:42.182438Z","iopub.status.idle":"2022-05-04T12:48:42.190225Z","shell.execute_reply.started":"2022-05-04T12:48:42.182403Z","shell.execute_reply":"2022-05-04T12:48:42.189478Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from sklearn.metrics import plot_confusion_matrix\nfrom sklearn.metrics import classification_report\nfrom cuml.metrics import confusion_matrix\nfrom sklearn.metrics import plot_confusion_matrix\nfrom sklearn.metrics import multilabel_confusion_matrix\n\nimport plotly.express as px","metadata":{"execution":{"iopub.status.busy":"2022-05-04T14:22:04.449225Z","iopub.execute_input":"2022-05-04T14:22:04.449484Z","iopub.status.idle":"2022-05-04T14:22:04.454732Z","shell.execute_reply.started":"2022-05-04T14:22:04.44945Z","shell.execute_reply":"2022-05-04T14:22:04.454017Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# _y_pred_  = list()\n# for item in y_pred:\n# #     print(item)\n#     _y_pred_.append(item)\n    \n# y_pred_ = np.array(_y_pred_, dtype=np.int32)\n# y_pred_[0]","metadata":{"execution":{"iopub.status.busy":"2022-05-04T12:50:31.416347Z","iopub.execute_input":"2022-05-04T12:50:31.416674Z","iopub.status.idle":"2022-05-04T12:50:31.424364Z","shell.execute_reply.started":"2022-05-04T12:50:31.416621Z","shell.execute_reply":"2022-05-04T12:50:31.423649Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"np.array(y_pred2)[0]","metadata":{"execution":{"iopub.status.busy":"2022-05-04T13:12:35.348687Z","iopub.execute_input":"2022-05-04T13:12:35.348944Z","iopub.status.idle":"2022-05-04T13:12:35.356698Z","shell.execute_reply.started":"2022-05-04T13:12:35.348916Z","shell.execute_reply":"2022-05-04T13:12:35.355701Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"for i in range(len(classes)):\n    print(\"[{}: {}]\".format(i, classes[i]), end=\", \")\n    \n\n# confM = list()\n# for i in range(len(svm_pca_experiment)):\nindex = 2\ni = svm_pca_experiment[index]\n\ny_pred_ = np.array(y_pred2, dtype=np.int32)\ny_test_ = np.array(y_test, dtype=np.int32)\n\nconf_matrix = cp.asnumpy(confusion_matrix(y_test_, y_pred_[0], normalize=\"true\"))\nconf_matrix= np.around(conf_matrix.astype('float') / conf_matrix.sum(axis=1)[:, np.newaxis], decimals=2)\n# conf_matrix = plot_confusion_matrix(model2, y_test_, y_pred_, normalize=\"true\", display_labels=classes)\n# conf_matrix = multilabel_confusion_matrix(np.array(y_test_), np.array(y_pred2)[0], labels=classes)\nconfM = conf_matrix\n\n\n# confM.append(conf_matrix)\n\nt = \"Normalized confusion matrix for SVM with \" + str(svm_pca_experiment[index][\"num_compon\"]) + \" PCA comp with \" + str(svm_pca_experiment[index][\"variance\"]) + \" variance\"\nfig = px.imshow(conf_matrix, text_auto=True, title=t, x=classes, y=classes)\n# fig.update_xaxes(dtick=classes)\nfig.show()\n\n# plt.title(t)\n# plt.show()\n# disp.confusion_matrix","metadata":{"execution":{"iopub.status.busy":"2022-05-04T14:25:14.161958Z","iopub.execute_input":"2022-05-04T14:25:14.162215Z","iopub.status.idle":"2022-05-04T14:25:14.224536Z","shell.execute_reply.started":"2022-05-04T14:25:14.162186Z","shell.execute_reply":"2022-05-04T14:25:14.223839Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# print(conf_matrix)\n# row1_sum = np.array(disp.confusion_matrix[0]).sum()\n# row2_sum = np.array(disp.confusion_matrix[1]).sum()\n\n# acc_Normal = (disp.confusion_matrix[0][0] / row1_sum)*100\n# acc_Tumor = (disp.confusion_matrix[1][1] / row2_sum)*100\n\nper_class_acc = list()\n\nfor i in range(len(classes)):\n    row_sum = np.array(confM[i]).sum()\n    acc = (conf_matrix[i][i]/row_sum)*100\n    per_class_acc.append(acc)\n    \nfor i in range(len(classes)):\n    print(\"[{}: {} - {}]\".format(i, classes[i], per_class_acc[i]), end=\", \")\n    ","metadata":{"execution":{"iopub.status.busy":"2022-05-03T20:23:38.481259Z","iopub.execute_input":"2022-05-03T20:23:38.482106Z","iopub.status.idle":"2022-05-03T20:23:38.491263Z","shell.execute_reply.started":"2022-05-03T20:23:38.482067Z","shell.execute_reply":"2022-05-03T20:23:38.490364Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from sklearn.metrics import roc_curve, auc\nfrom matplotlib.pyplot import figure","metadata":{"execution":{"iopub.status.busy":"2022-05-03T20:23:38.493072Z","iopub.execute_input":"2022-05-03T20:23:38.493381Z","iopub.status.idle":"2022-05-03T20:23:38.504393Z","shell.execute_reply.started":"2022-05-03T20:23:38.493346Z","shell.execute_reply":"2022-05-03T20:23:38.503502Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"model2.get_params()","metadata":{"execution":{"iopub.status.busy":"2022-05-03T20:23:38.505684Z","iopub.execute_input":"2022-05-03T20:23:38.50635Z","iopub.status.idle":"2022-05-03T20:23:38.516664Z","shell.execute_reply.started":"2022-05-03T20:23:38.506295Z","shell.execute_reply":"2022-05-03T20:23:38.515885Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from sklearn.preprocessing import LabelBinarizer","metadata":{"execution":{"iopub.status.busy":"2022-05-03T20:23:38.518071Z","iopub.execute_input":"2022-05-03T20:23:38.518626Z","iopub.status.idle":"2022-05-03T20:23:38.526242Z","shell.execute_reply.started":"2022-05-03T20:23:38.518586Z","shell.execute_reply":"2022-05-03T20:23:38.525294Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"type(y_test_)","metadata":{"execution":{"iopub.status.busy":"2022-05-03T20:23:38.527633Z","iopub.execute_input":"2022-05-03T20:23:38.528415Z","iopub.status.idle":"2022-05-03T20:23:38.536495Z","shell.execute_reply.started":"2022-05-03T20:23:38.528374Z","shell.execute_reply":"2022-05-03T20:23:38.535594Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"temp = list()\nfor item in y_pred2[0]:\n    temp.append(int(item))","metadata":{"execution":{"iopub.status.busy":"2022-05-03T20:23:38.538135Z","iopub.execute_input":"2022-05-03T20:23:38.538709Z","iopub.status.idle":"2022-05-03T20:23:38.549041Z","shell.execute_reply.started":"2022-05-03T20:23:38.538671Z","shell.execute_reply":"2022-05-03T20:23:38.548151Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"y_pred2 = np.array(y_pred2).astype(int)","metadata":{"execution":{"iopub.status.busy":"2022-05-03T20:23:38.550763Z","iopub.execute_input":"2022-05-03T20:23:38.551067Z","iopub.status.idle":"2022-05-03T20:23:38.558399Z","shell.execute_reply.started":"2022-05-03T20:23:38.551012Z","shell.execute_reply":"2022-05-03T20:23:38.557483Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"lb = LabelBinarizer()\n# temp = np.array(y_pred2[0])\nenc_y_pred2 = lb.fit_transform(y_pred2[0])\nenc_y_test_ = lb.fit_transform(y_test_)","metadata":{"execution":{"iopub.status.busy":"2022-05-03T20:23:38.559964Z","iopub.execute_input":"2022-05-03T20:23:38.560334Z","iopub.status.idle":"2022-05-03T20:23:38.571462Z","shell.execute_reply.started":"2022-05-03T20:23:38.560283Z","shell.execute_reply":"2022-05-03T20:23:38.570516Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Compute ROC curve and ROC area for each class\nfpr = dict()\ntpr = dict()\nroc_auc = dict()\nfor i in range(len(classes)):\n    fpr[i], tpr[i], _ = roc_curve(enc_y_test_[:,i], enc_y_pred2[:, i])\n    roc_auc[i] = auc(fpr[i], tpr[i])","metadata":{"execution":{"iopub.status.busy":"2022-05-03T20:23:38.572714Z","iopub.execute_input":"2022-05-03T20:23:38.573449Z","iopub.status.idle":"2022-05-03T20:23:38.587867Z","shell.execute_reply.started":"2022-05-03T20:23:38.573397Z","shell.execute_reply":"2022-05-03T20:23:38.587128Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Compute micro-average ROC curve and ROC area\nfpr[\"micro\"], tpr[\"micro\"], _ = roc_curve(enc_y_test_.ravel(), enc_y_pred2.ravel())\nroc_auc[\"micro\"] = auc(fpr[\"micro\"], tpr[\"micro\"])","metadata":{"execution":{"iopub.status.busy":"2022-05-03T20:23:38.589071Z","iopub.execute_input":"2022-05-03T20:23:38.589831Z","iopub.status.idle":"2022-05-03T20:23:38.600095Z","shell.execute_reply.started":"2022-05-03T20:23:38.589795Z","shell.execute_reply":"2022-05-03T20:23:38.598942Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Plot ROC curve\nplt.figure()\nplt.plot(fpr[\"micro\"], tpr[\"micro\"],\n         label='micro-average ROC curve (area = {0:0.2f})'\n               ''.format(roc_auc[\"micro\"]))\nfor i in range(len(classes)):\n    plt.plot(fpr[i], tpr[i], label='ROC curve of class {0} (area = {1:0.2f})'\n                                   ''.format(i, roc_auc[i]))\n\n# figure(figsize=(8, 6), dpi=80)\nplt.plot([0, 1], [0, 1], 'k--')\nplt.xlim([0.0, 1.0])\nplt.ylim([0.0, 1.05])\nplt.xlabel('False Positive Rate')\nplt.ylabel('True Positive Rate')\nplt.title('Some extension of Receiver operating characteristic to multi-class')\nplt.legend(bbox_to_anchor=(1.04,1), borderaxespad=0)\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2022-05-03T20:23:38.601913Z","iopub.execute_input":"2022-05-03T20:23:38.602357Z","iopub.status.idle":"2022-05-03T20:23:39.009754Z","shell.execute_reply.started":"2022-05-03T20:23:38.602319Z","shell.execute_reply":"2022-05-03T20:23:39.009092Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# classes\n\n# for i in range(len(svm_pca_experiment)):\nprint(\"Classification report for SVM with \" + str(svm_pca_experiment[i][\"num_compon\"]) + \" PCA comp with \" + str(svm_pca_experiment[i][\"variance\"]) + \" variance\")\ny_pred_ = y_pred2[0]\nprint(classification_report(y_test, y_pred_, target_names=classes))","metadata":{"execution":{"iopub.status.busy":"2022-05-03T20:23:39.011023Z","iopub.execute_input":"2022-05-03T20:23:39.011506Z","iopub.status.idle":"2022-05-03T20:23:39.032769Z","shell.execute_reply.started":"2022-05-03T20:23:39.011467Z","shell.execute_reply":"2022-05-03T20:23:39.031985Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}