{"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":"# Bussiness Task\nThe goal of this competition is to classify the blood clot origins in ischemic stroke. Using whole slide digital pathology images, i'll build a model that differentiates between the two major acute ischemic stroke (AIS) etiology subtypes: cardiac and large artery atherosclerosis.\n\n\nI hope you find this NoteBook helpful and some **UPVOTES** would be appreciated.\n","metadata":{}},{"cell_type":"markdown","source":"# what is ischemic stroke ?\n\n","metadata":{}},{"cell_type":"markdown","source":"","metadata":{}},{"cell_type":"markdown","source":"An ischaemic stroke happens when a blockage cuts off the blood supply to part of your brain, killing brain cells. Damage to brain cells can affect how the body works. It can also change how you think and feel.\n\nIt’s the most common type of stroke, and around 85% of strokes in the UK are ischaemic strokes. The other 15% of strokes are due to bleeding in or around the brain, known as haemorrhagic stroke. \n\nA transient ischaemic attack (TIA or mini-stroke) is the same as a stroke but the symptoms only last for a short amount of time. It is a major warning sign of a stroke and should always be taken seriously.","metadata":{}},{"cell_type":"markdown","source":"\n\n# Import libraries.\n","metadata":{"id":"zqrYqv734KAo"}},{"cell_type":"code","source":"# Core\nimport numpy as np\nimport pandas as pd \nimport matplotlib.pyplot as plt\nfrom pathlib import Path\nimport seaborn as sns\nimport glob\nimport random\nimport os\nimport cv2\nimport gc\nfrom tqdm import tqdm\n#tesor fow & keras\nimport tensorflow as tf\nfrom tensorflow import keras\nfrom keras.regularizers import l2 \nfrom sklearn.utils import shuffle\nfrom sklearn.model_selection import train_test_split\nfrom tensorflow.keras.utils import to_categorical\nfrom keras.models import load_model\nfrom tensorflow.keras import Sequential\nfrom tensorflow.keras.layers import Flatten, Dense,BatchNormalization,Dropout,Input\nfrom keras.models import Sequential, Model\nfrom keras.layers import Conv2D,GlobalMaxPooling2D\nfrom tensorflow.keras.applications import Xception,VGG16,InceptionResNetV2\nfrom tensorflow.keras.preprocessing.image import ImageDataGenerator\n#cnn\nfrom tensorflow.keras import datasets, layers, models\n\nfrom keras.layers import Input, Conv2D, MaxPooling2D, UpSampling2D, concatenate, Conv2DTranspose, BatchNormalization, Dropout, Lambda\nfrom keras.engine.base_layer import Layer\nfrom sklearn.metrics import classification_report,confusion_matrix\nfrom sklearn.metrics import accuracy_score\nfrom sklearn.metrics import f1_score\nfrom sklearn.metrics import recall_score\nfrom sklearn.metrics import precision_score\nfrom sklearn.utils import class_weight\nfrom sklearn.utils.class_weight import compute_class_weight\nfrom PIL import Image\nImage.MAX_IMAGE_PIXELS = None\nimport PIL\nimport tifffile\nimport matplotlib.pyplot as plt\ngc.enable()","metadata":{"id":"1eb99db7","execution":{"iopub.status.busy":"2022-08-14T21:19:39.288512Z","iopub.execute_input":"2022-08-14T21:19:39.289019Z","iopub.status.idle":"2022-08-14T21:19:47.309172Z","shell.execute_reply.started":"2022-08-14T21:19:39.288921Z","shell.execute_reply":"2022-08-14T21:19:47.307945Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"seed = 42\nnp.random.seed =seed","metadata":{"id":"4DTeuvL_TphS","execution":{"iopub.status.busy":"2022-08-14T21:19:47.311189Z","iopub.execute_input":"2022-08-14T21:19:47.311833Z","iopub.status.idle":"2022-08-14T21:19:47.322385Z","shell.execute_reply.started":"2022-08-14T21:19:47.311798Z","shell.execute_reply":"2022-08-14T21:19:47.317855Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"image_path = '../input/mayo-clinic-strip-ai/'","metadata":{"id":"qcaudVrD0uQ2","execution":{"iopub.status.busy":"2022-08-14T21:19:47.323819Z","iopub.execute_input":"2022-08-14T21:19:47.324248Z","iopub.status.idle":"2022-08-14T21:19:47.337273Z","shell.execute_reply.started":"2022-08-14T21:19:47.324213Z","shell.execute_reply":"2022-08-14T21:19:47.336168Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Helper Function\n\n\n","metadata":{"id":"2zDVR6bc8DyI"}},{"cell_type":"code","source":"#check duplicate data \ndef check_duplicate(df):\n if df.duplicated().all():\n return 'There are duplicate Data in Data Frame Nedded To be removed ' \n else :\n return 'Data Is clean ,No Duplicate Data Found '\n\n# get label Name \ndef get_Label(number):\n labels = {0:'CE', 1:'LAA'}\n return labels[number]\n\n\n#plot predction function\ndef plot_predection(model_name):\n plt.figure(figsize=(20,15))\n plt.suptitle(\"Predection Images\", fontsize=20)\n images = [] \n path =image_path+'/'+'test1/'\n count = 0 #val_images,val_labels\n for i,files in enumerate(os.listdir(path)) :\n img = plt.imread(path+files)\n img = cv2.resize(img,(128,128))\n plt.imshow(img,cmap=plt.cm.binary)\n img = np.expand_dims(img, axis=0)\n feature = model_name.predict(img)\n predection = np.argmax(feature, axis=1)\n print(predection)\n plt.subplot(5,7,i+1)\n plt.xticks([])\n plt.yticks([])\n plt.grid(False)\n plt.xlabel(\"Predicted\"+get_Label(int(predection)))\n # plt.xlabel(\"Predicted \"+ int(predection))\n\n # plt.ylabel(get_Label(val_labels[i]))\n count += 1\n if count == 34 :\n break \ndef get_id (x):\n id = x.split('.')[0].split('_0')[0]\n return id \n\ndef freezing_layers(model_name):\n for layer in model_name.layers:\n layer.trainable = False \n \ndef numerical_plotting(df, col, title, symb):\n fig, ax = plt.subplots(2, 1, sharex=True, figsize=(8,5),gridspec_kw={\"height_ratios\": (.2, .8)})\n ax[0].set_title(title,fontsize=18)\n sns.boxplot(x=col, data=df, ax=ax[0])\n ax[0].set(yticks=[])\n sns.distplot(df[col],kde=True)\n plt.xticks(rotation=45)\n ax[1].set_xlabel(col, fontsize=16)\n plt.axvline(df[col].mean(), color='darkgreen', linewidth=2.2, label='mean=' + str(np.round(df[col].mean(),1)) + symb)\n plt.axvline(df[col].median(), color='red', linewidth=2.2, label='median='+ str(np.round(df[col].median(),1)) + symb)\n plt.axvline(df[col].mode()[0], color='purple', linewidth=2.2, label='mode='+ str(df[col].mode()[0]) + symb)\n plt.legend(bbox_to_anchor=(1, 1.03), ncol=1, fontsize=17, fancybox=True, shadow=True, frameon=True)\n plt.tight_layout()\n plt.show() \n\ndef categorical_plotting(df,col,title):\n fig, ax = plt.subplots(figsize=(10,5))\n ax=sns.countplot(x=col, data=df, palette='flare', order = df[col].value_counts().index)\n ax.set_xticklabels(ax.get_xticklabels(), rotation=45)\n ax.bar_label(ax.containers[0])\n plt.title(title)\n plt.show()\n \ndef average_plotting(df,col,output,number,title):\n data_list = df[col].value_counts().index[:number].tolist()\n plt.figure(figsize=(15,5))\n ax=sns.barplot(x=col, y=output, data=df[df[col].isin(data_list)],order=data_list,palette='flare',ci=False,edgecolor=\"black\") \n plt.xticks(rotation=45);\n ax.bar_label(ax.containers[0])\n plt.title(title)\n plt.show() \n \ndef draw_unique_value (df,title):\n plt.figure(figsize=(10,5))\n plt.title(title)\n unique_counts = df.nunique().to_dict()\n ax = sns.barplot(list(unique_counts.keys()), list(unique_counts.values()),palette='flare')\n ax.bar_label(ax.containers[0])\n plt.plot()","metadata":{"id":"veeEUr6oElre","execution":{"iopub.status.busy":"2022-08-14T21:19:47.340518Z","iopub.execute_input":"2022-08-14T21:19:47.341604Z","iopub.status.idle":"2022-08-14T21:19:47.367924Z","shell.execute_reply.started":"2022-08-14T21:19:47.341543Z","shell.execute_reply":"2022-08-14T21:19:47.366976Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Loading Preperation ","metadata":{"id":"5ssSyiLDdTlP"}},{"cell_type":"code","source":"train_df = pd.read_csv(image_path + 'train.csv')\ntest_df = pd.read_csv(image_path + 'test.csv')\nother_df = pd.read_csv(image_path + 'other.csv')","metadata":{"id":"t5Acfd4xik4n","outputId":"d7014f41-eca0-41da-b840-a3c81f6ee027","execution":{"iopub.status.busy":"2022-08-14T21:19:47.369309Z","iopub.execute_input":"2022-08-14T21:19:47.369942Z","iopub.status.idle":"2022-08-14T21:19:47.40707Z","shell.execute_reply.started":"2022-08-14T21:19:47.369885Z","shell.execute_reply":"2022-08-14T21:19:47.405957Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Data Analaysis & Visualization\nin this part we will analays and versialize each part of data to be in near step from our goal then pased on deployed models we will sense best factior that affect on our bussiness goal","metadata":{}},{"cell_type":"code","source":"train_df.head(5)","metadata":{"id":"jT6svNwoq6cA","outputId":"cf363dbf-f182-448f-b6a0-7ecb67eddb1b","execution":{"iopub.status.busy":"2022-08-14T21:19:47.411078Z","iopub.execute_input":"2022-08-14T21:19:47.411444Z","iopub.status.idle":"2022-08-14T21:19:47.436995Z","shell.execute_reply.started":"2022-08-14T21:19:47.411411Z","shell.execute_reply":"2022-08-14T21:19:47.435985Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# store path of images\ntrain_images_paths = []\ntrain_images =[]\ntrain_path =image_path+'train/'\nfor i,img in enumerate(os.listdir(train_path)) :\n image_full_path = train_path+img\n train_images_paths.append(image_full_path)\n train_images.append(img)\n # add path to data frame\ntrain_df['path'] = train_images_paths \ntrain_df['image'] = train_images \ntrain_df.head(5)","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:47.438424Z","iopub.execute_input":"2022-08-14T21:19:47.438745Z","iopub.status.idle":"2022-08-14T21:19:47.526326Z","shell.execute_reply.started":"2022-08-14T21:19:47.438715Z","shell.execute_reply":"2022-08-14T21:19:47.525399Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# store path of test images\ntest_images_paths = []\ntest_images =[]\ntest_path =image_path+'test/'\nfor i,img in enumerate(os.listdir(test_path)) :\n image_full_path = test_path+img\n test_images_paths.append(image_full_path)\n test_images.append(img)\n # add path to test data frame\ntest_df['path'] = test_images_paths \ntest_df['image'] = test_images \n\ntest_df.head(5)","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:47.527715Z","iopub.execute_input":"2022-08-14T21:19:47.528736Z","iopub.status.idle":"2022-08-14T21:19:47.548787Z","shell.execute_reply.started":"2022-08-14T21:19:47.528687Z","shell.execute_reply":"2022-08-14T21:19:47.547346Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_df.info()","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:47.550267Z","iopub.execute_input":"2022-08-14T21:19:47.550887Z","iopub.status.idle":"2022-08-14T21:19:47.578954Z","shell.execute_reply.started":"2022-08-14T21:19:47.550852Z","shell.execute_reply":"2022-08-14T21:19:47.577722Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_df.shape","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:47.583832Z","iopub.execute_input":"2022-08-14T21:19:47.584994Z","iopub.status.idle":"2022-08-14T21:19:47.59629Z","shell.execute_reply.started":"2022-08-14T21:19:47.584951Z","shell.execute_reply":"2022-08-14T21:19:47.5947Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# check null value\ntrain_df.isnull().sum().sort_values(ascending=False)","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:47.597997Z","iopub.execute_input":"2022-08-14T21:19:47.599197Z","iopub.status.idle":"2022-08-14T21:19:47.611542Z","shell.execute_reply.started":"2022-08-14T21:19:47.599138Z","shell.execute_reply":"2022-08-14T21:19:47.610508Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#check duplicate data \ncheck_duplicate(train_df)","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:47.613125Z","iopub.execute_input":"2022-08-14T21:19:47.616203Z","iopub.status.idle":"2022-08-14T21:19:47.628846Z","shell.execute_reply.started":"2022-08-14T21:19:47.615863Z","shell.execute_reply":"2022-08-14T21:19:47.627971Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"cols =train_df.columns\nsns.heatmap(train_df[cols].isnull(), cmap='viridis')","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:47.630466Z","iopub.execute_input":"2022-08-14T21:19:47.630931Z","iopub.status.idle":"2022-08-14T21:19:48.019386Z","shell.execute_reply.started":"2022-08-14T21:19:47.63086Z","shell.execute_reply":"2022-08-14T21:19:48.018215Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_df['image_id'].value_counts()","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:48.020998Z","iopub.execute_input":"2022-08-14T21:19:48.021363Z","iopub.status.idle":"2022-08-14T21:19:48.034992Z","shell.execute_reply.started":"2022-08-14T21:19:48.02133Z","shell.execute_reply":"2022-08-14T21:19:48.033827Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_df.info()","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:48.036682Z","iopub.execute_input":"2022-08-14T21:19:48.037062Z","iopub.status.idle":"2022-08-14T21:19:48.049409Z","shell.execute_reply.started":"2022-08-14T21:19:48.03703Z","shell.execute_reply":"2022-08-14T21:19:48.048191Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#get numerical column\nnumCol = [col for col in train_df.columns if train_df[col].dtype != \"O\"]\nnumCol","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:48.050862Z","iopub.execute_input":"2022-08-14T21:19:48.051539Z","iopub.status.idle":"2022-08-14T21:19:48.058882Z","shell.execute_reply.started":"2022-08-14T21:19:48.051503Z","shell.execute_reply":"2022-08-14T21:19:48.057853Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#get categoriacl column\nCatCol = [col for col in train_df.columns if train_df[col].dtype == \"O\"]\nCatCol","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:48.060509Z","iopub.execute_input":"2022-08-14T21:19:48.061379Z","iopub.status.idle":"2022-08-14T21:19:48.071512Z","shell.execute_reply.started":"2022-08-14T21:19:48.061331Z","shell.execute_reply":"2022-08-14T21:19:48.070294Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":" draw_unique_value (train_df,'Unique Value Of Data Frame')","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:48.072962Z","iopub.execute_input":"2022-08-14T21:19:48.073605Z","iopub.status.idle":"2022-08-14T21:19:48.315639Z","shell.execute_reply.started":"2022-08-14T21:19:48.073564Z","shell.execute_reply":"2022-08-14T21:19:48.314751Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"**Center_id** \n* **Identifies the medical center where the slide was obtained** .","metadata":{}},{"cell_type":"code","source":"numerical_plotting(train_df,'center_id','Centar id Distribution',' ')","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:48.317114Z","iopub.execute_input":"2022-08-14T21:19:48.317706Z","iopub.status.idle":"2022-08-14T21:19:48.765006Z","shell.execute_reply.started":"2022-08-14T21:19:48.31766Z","shell.execute_reply":"2022-08-14T21:19:48.764146Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_df['center_id'].value_counts()","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:48.766381Z","iopub.execute_input":"2022-08-14T21:19:48.76697Z","iopub.status.idle":"2022-08-14T21:19:48.774816Z","shell.execute_reply.started":"2022-08-14T21:19:48.766934Z","shell.execute_reply":"2022-08-14T21:19:48.773801Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_df.describe()[['center_id']].T","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:48.776623Z","iopub.execute_input":"2022-08-14T21:19:48.777019Z","iopub.status.idle":"2022-08-14T21:19:48.800781Z","shell.execute_reply.started":"2022-08-14T21:19:48.776976Z","shell.execute_reply":"2022-08-14T21:19:48.799765Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#what is the most top center_id in data set ?\ntrain_df['center_id'].sort_values(ascending=False).head(5)","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:48.802353Z","iopub.execute_input":"2022-08-14T21:19:48.802693Z","iopub.status.idle":"2022-08-14T21:19:48.811398Z","shell.execute_reply.started":"2022-08-14T21:19:48.802655Z","shell.execute_reply":"2022-08-14T21:19:48.810519Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"**Notes** ⏳\n\n* **max center id is 11 & min is 1.**\n* **there are 11 center which collect our data .**","metadata":{}},{"cell_type":"markdown","source":"**Patient_id** ","metadata":{}},{"cell_type":"code","source":"train_df['patient_id'].value_counts()","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:48.813076Z","iopub.execute_input":"2022-08-14T21:19:48.813684Z","iopub.status.idle":"2022-08-14T21:19:48.822419Z","shell.execute_reply.started":"2022-08-14T21:19:48.813642Z","shell.execute_reply":"2022-08-14T21:19:48.821451Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"categorical_plotting(train_df,'patient_id','total count of patient_id per Image')","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:48.823952Z","iopub.execute_input":"2022-08-14T21:19:48.824493Z","iopub.status.idle":"2022-08-14T21:19:57.944617Z","shell.execute_reply.started":"2022-08-14T21:19:48.824451Z","shell.execute_reply":"2022-08-14T21:19:57.943367Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"average_plotting(train_df,'patient_id','center_id',15,'Average Cental_id per Patient')","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:57.946237Z","iopub.execute_input":"2022-08-14T21:19:57.946661Z","iopub.status.idle":"2022-08-14T21:19:58.642288Z","shell.execute_reply.started":"2022-08-14T21:19:57.946623Z","shell.execute_reply":"2022-08-14T21:19:58.64085Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#what is total number of images per patient id ?\ntrain_df.groupby('patient_id').count()[['image_num']].sort_values(by='image_num',ascending=False).head(5).T","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:58.644161Z","iopub.execute_input":"2022-08-14T21:19:58.644638Z","iopub.status.idle":"2022-08-14T21:19:58.662297Z","shell.execute_reply.started":"2022-08-14T21:19:58.644589Z","shell.execute_reply":"2022-08-14T21:19:58.661206Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_df[train_df['patient_id'] == '91b9d3']","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:58.664938Z","iopub.execute_input":"2022-08-14T21:19:58.665996Z","iopub.status.idle":"2022-08-14T21:19:58.682325Z","shell.execute_reply.started":"2022-08-14T21:19:58.665943Z","shell.execute_reply":"2022-08-14T21:19:58.681284Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"##what is total of images per patient id ?\ntrain_df.groupby('patient_id').sum()[['image_num']].sort_values(by='image_num',ascending=False).head(5).T","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:58.690194Z","iopub.execute_input":"2022-08-14T21:19:58.690573Z","iopub.status.idle":"2022-08-14T21:19:58.707775Z","shell.execute_reply.started":"2022-08-14T21:19:58.69054Z","shell.execute_reply":"2022-08-14T21:19:58.706621Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"**image_num** ","metadata":{}},{"cell_type":"code","source":"train_df['image_num'].value_counts()","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:58.709238Z","iopub.execute_input":"2022-08-14T21:19:58.709665Z","iopub.status.idle":"2022-08-14T21:19:58.721143Z","shell.execute_reply.started":"2022-08-14T21:19:58.709619Z","shell.execute_reply":"2022-08-14T21:19:58.719827Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"categorical_plotting(train_df,'image_num','total count of image per Paient')","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:58.722677Z","iopub.execute_input":"2022-08-14T21:19:58.723693Z","iopub.status.idle":"2022-08-14T21:19:58.92629Z","shell.execute_reply.started":"2022-08-14T21:19:58.723654Z","shell.execute_reply":"2022-08-14T21:19:58.925153Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":" \n**Notes** ⏳\n\n* **most of patient has one image**","metadata":{}},{"cell_type":"code","source":"#what is max image per patient ?\ntrain_df.groupby(['patient_id']).sum()[['image_num']].sort_values(by ='image_num',ascending=False).head(5).T","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:58.92778Z","iopub.execute_input":"2022-08-14T21:19:58.928536Z","iopub.status.idle":"2022-08-14T21:19:58.945692Z","shell.execute_reply.started":"2022-08-14T21:19:58.928489Z","shell.execute_reply":"2022-08-14T21:19:58.944439Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"**label** ","metadata":{}},{"cell_type":"code","source":"train_df['label'].value_counts()","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:58.947304Z","iopub.execute_input":"2022-08-14T21:19:58.947743Z","iopub.status.idle":"2022-08-14T21:19:58.95615Z","shell.execute_reply.started":"2022-08-14T21:19:58.947699Z","shell.execute_reply":"2022-08-14T21:19:58.955086Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"categorical_plotting(train_df,'label','total count of image per label')","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:58.9577Z","iopub.execute_input":"2022-08-14T21:19:58.95806Z","iopub.status.idle":"2022-08-14T21:19:59.131656Z","shell.execute_reply.started":"2022-08-14T21:19:58.958028Z","shell.execute_reply":"2022-08-14T21:19:59.130755Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"* **from above we found that data is not balanced** .\n* **72.5 % of data belong tom class CE .**\n","metadata":{}},{"cell_type":"markdown","source":"# Visialize Training Images","metadata":{}},{"cell_type":"code","source":"# plt.figure(figsize=(20,15))\n# plt.suptitle(\"Training Images\", fontsize=20)\n# path =image_path+'train/'\n# counter =0\n# for i,img in enumerate(os.listdir(path)) :\n# plt.subplot(3,3,i+1)\n# full_image= Image.open(path+img)\n# full_image_1 =full_image.resize((512,512))\n# plt.xticks([])\n# plt.yticks([])\n# plt.grid(False)\n# plt.imshow(full_image_1, cmap=plt.cm.binary) \n# if i == 8:\n# break\n# del full_image_1\n# gc.collect()","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:59.133158Z","iopub.execute_input":"2022-08-14T21:19:59.133759Z","iopub.status.idle":"2022-08-14T21:19:59.137842Z","shell.execute_reply.started":"2022-08-14T21:19:59.133726Z","shell.execute_reply":"2022-08-14T21:19:59.137026Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Image Resizing\n* Due to hardware limitation i will convert all image size to 256 to contiune to reach compition Goal .","metadata":{}},{"cell_type":"code","source":" #change image path\nnew_image_path = 'kaggle/working/'","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:59.139057Z","iopub.execute_input":"2022-08-14T21:19:59.139854Z","iopub.status.idle":"2022-08-14T21:19:59.150921Z","shell.execute_reply.started":"2022-08-14T21:19:59.139822Z","shell.execute_reply":"2022-08-14T21:19:59.149803Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#create ne path to store images \ncreate_train_path_new =os.makedirs(new_image_path+'train')\ncreate_test_path_new =os.makedirs(new_image_path+'test')","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:59.152245Z","iopub.execute_input":"2022-08-14T21:19:59.152626Z","iopub.status.idle":"2022-08-14T21:19:59.16385Z","shell.execute_reply.started":"2022-08-14T21:19:59.152582Z","shell.execute_reply":"2022-08-14T21:19:59.162687Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_path_new = new_image_path+'train'\ntest_path_new = new_image_path+'test'","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:59.165521Z","iopub.execute_input":"2022-08-14T21:19:59.165925Z","iopub.status.idle":"2022-08-14T21:19:59.173606Z","shell.execute_reply.started":"2022-08-14T21:19:59.165869Z","shell.execute_reply":"2022-08-14T21:19:59.172576Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"os.listdir('kaggle/working')","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:59.175228Z","iopub.execute_input":"2022-08-14T21:19:59.175609Z","iopub.status.idle":"2022-08-14T21:19:59.187706Z","shell.execute_reply.started":"2022-08-14T21:19:59.175573Z","shell.execute_reply":"2022-08-14T21:19:59.186868Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"\ntrain_images_paths_new = []\ntrain_images_new = []\ntrain_path =image_path+'train/'\nfor img in tqdm(os.listdir(train_path)):\n image_full_path = train_path+img\n image_new_name = image_full_path.split('/')[-1].split('.')[0]\n imge_new_pathe= train_path_new+'/'+image_new_name+'.jpg'\n train_images_paths_new.append(imge_new_pathe)\n train_images_new.append(image_new_name+'.jpg')\n image = tifffile.imread(image_full_path)\n resized_img = cv2.resize(image,(256,256))\n cv2.imwrite(train_path_new+'/'+image_new_name+'.jpg', resized_img)\n del image_full_path\n del image_new_name\n del image\n del resized_img\n gc.collect()\n \ntrain_df['new_path'] = train_images_paths_new\ntrain_df['new_image_name'] = train_images_new","metadata":{"execution":{"iopub.status.busy":"2022-08-14T21:19:59.189113Z","iopub.execute_input":"2022-08-14T21:19:59.190682Z","iopub.status.idle":"2022-08-15T00:35:48.20107Z","shell.execute_reply.started":"2022-08-14T21:19:59.190629Z","shell.execute_reply":"2022-08-15T00:35:48.199325Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"test_images_paths_new = []\ntest_images_new = []\ntest_path =image_path+'test/'\n\nfor img in tqdm(os.listdir(test_path)):\n image_full_path = test_path+img\n image_new_name = image_full_path.split('/')[-1].split('.')[0]\n imge_new_pathe= test_path_new+'/'+image_new_name+'.jpg'\n test_images_paths_new.append(imge_new_pathe)\n test_images_new.append(image_new_name+'.jpg')\n image = tifffile.imread(image_full_path)\n resized_img = cv2.resize(image,(256,256))\n cv2.imwrite(test_path_new+'/'+image_new_name+'.jpg', resized_img)\n del image_full_path\n del image_new_name\n del image\n del resized_img\n gc.collect()\n \ntest_df['new_path'] = test_images_paths_new\ntest_df['new_image_name'] = test_images_new\n","metadata":{"execution":{"iopub.status.busy":"2022-08-15T00:35:48.20353Z","iopub.execute_input":"2022-08-15T00:35:48.20512Z","iopub.status.idle":"2022-08-15T00:37:14.236307Z","shell.execute_reply.started":"2022-08-15T00:35:48.205065Z","shell.execute_reply":"2022-08-15T00:37:14.234994Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#create new path for train and test data to start modiling\ntrain_path_new = new_image_path+'train'\ntest_path_new = new_image_path+'test'","metadata":{"execution":{"iopub.status.busy":"2022-08-15T00:37:14.23824Z","iopub.execute_input":"2022-08-15T00:37:14.239046Z","iopub.status.idle":"2022-08-15T00:37:14.244576Z","shell.execute_reply.started":"2022-08-15T00:37:14.238994Z","shell.execute_reply":"2022-08-15T00:37:14.243679Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# import shutil\n# os.chdir(r'/kaggle/working')\n# shutil.make_archive('images', 'zip', 'kaggle/working')","metadata":{"execution":{"iopub.status.busy":"2022-08-15T00:37:14.24558Z","iopub.execute_input":"2022-08-15T00:37:14.245975Z","iopub.status.idle":"2022-08-15T00:37:14.259515Z","shell.execute_reply.started":"2022-08-15T00:37:14.24593Z","shell.execute_reply":"2022-08-15T00:37:14.258141Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# from IPython.display import FileLink\n# FileLink(r'./images.zip')","metadata":{"execution":{"iopub.status.busy":"2022-08-15T00:37:14.261251Z","iopub.execute_input":"2022-08-15T00:37:14.261857Z","iopub.status.idle":"2022-08-15T00:37:14.273826Z","shell.execute_reply.started":"2022-08-15T00:37:14.261819Z","shell.execute_reply":"2022-08-15T00:37:14.271767Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_df.head(5)","metadata":{"execution":{"iopub.status.busy":"2022-08-15T00:37:14.276586Z","iopub.execute_input":"2022-08-15T00:37:14.277216Z","iopub.status.idle":"2022-08-15T00:37:14.30424Z","shell.execute_reply.started":"2022-08-15T00:37:14.277163Z","shell.execute_reply":"2022-08-15T00:37:14.302991Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Buliding CNN Model\n\n**i will use Data augmentation,as it is a set of techniques to artificially increase the amount of data by generating new data points from existing data. This includes making small changes to data or using deep learning models to generate new data points.** \n\n","metadata":{"id":"5O1NZpjrijnM"}},{"cell_type":"code","source":"batch_size = 8\n#create image generator for images \nimage_gen = ImageDataGenerator(\n rescale = 1./255,\n shear_range = 0.2,\n zoom_range = 0.5,\n height_shift_range=0.2,\n width_shift_range=0.2,\n fill_mode='nearest',\n horizontal_flip=True,\n rotation_range = 20,\n validation_split=0.2 \n )\n\n \ntrain = image_gen.flow_from_dataframe(\n train_df,\n train_path_new+'/',\n x_col='new_image_name',\n y_col='label',\n target_size=(96,96),\n class_mode='categorical',\n shuffle=True, \n batch_size=batch_size,\n subset = \"training\",\n color_mode='rgb',\n )\nvalidate = image_gen.flow_from_dataframe(\n train_df,\n train_path_new+'/',\n x_col='new_image_name',\n y_col='label',\n target_size=(96,96),\n class_mode='categorical',\n shuffle=True, \n batch_size=batch_size,\n subset = \"validation\",\n color_mode='rgb'\n )\n\n\n# make labels is balanced\nclass_weights = compute_class_weight(\n class_weight = \"balanced\",\n classes = np.unique(train.classes),\n y = train.classes \n )\nclass_weights = dict(zip(np.unique(train.classes), class_weights))\nclass_weights\n","metadata":{"id":"p7jQbFP742OG","outputId":"42dbe29c-86fa-421c-d95f-b9dc34852c21","execution":{"iopub.status.busy":"2022-08-15T00:37:14.306039Z","iopub.execute_input":"2022-08-15T00:37:14.306423Z","iopub.status.idle":"2022-08-15T00:37:14.365316Z","shell.execute_reply.started":"2022-08-15T00:37:14.306388Z","shell.execute_reply":"2022-08-15T00:37:14.363949Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train.class_indices","metadata":{"execution":{"iopub.status.busy":"2022-08-15T00:37:14.36705Z","iopub.execute_input":"2022-08-15T00:37:14.367776Z","iopub.status.idle":"2022-08-15T00:37:14.375391Z","shell.execute_reply.started":"2022-08-15T00:37:14.367736Z","shell.execute_reply":"2022-08-15T00:37:14.37418Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"cnn_model = models.Sequential()\ncnn_model.add(layers.Conv2D(64,(5,5), kernel_regularizer=l2(0.00005),padding ='Same',activation = 'relu',input_shape=(96,96,3)))\ncnn_model.add(layers.MaxPooling2D(2,2))\ncnn_model.add(BatchNormalization())\ncnn_model.add(layers.Conv2D(128,(5,5) ,kernel_regularizer=l2(0.00005),padding ='same',activation='relu'))\ncnn_model.add(layers.MaxPooling2D(2,2))\ncnn_model.add(BatchNormalization())\ncnn_model.add(layers.Conv2D(256,(5,5) ,kernel_regularizer=l2(0.00005),padding ='same',activation='relu'))\ncnn_model.add(layers.MaxPooling2D(2,2))\ncnn_model.add(BatchNormalization())\ncnn_model.summary()","metadata":{"id":"QfsDiDZE42Qz","outputId":"17e7b3d0-b9dc-458a-a5fa-ec1e50f2dd64","execution":{"iopub.status.busy":"2022-08-15T00:37:14.376594Z","iopub.execute_input":"2022-08-15T00:37:14.377035Z","iopub.status.idle":"2022-08-15T00:37:14.827789Z","shell.execute_reply.started":"2022-08-15T00:37:14.376999Z","shell.execute_reply":"2022-08-15T00:37:14.825673Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"cnn_model.add(layers.Flatten())\ncnn_model.add(layers.Dense(128, activation='relu'))\ncnn_model.add(BatchNormalization())\ncnn_model.add(Dropout(0.5))\n# cnn_model.add(layers.Dense(64, activation='relu'))\n# cnn_model.add(BatchNormalization())\n# cnn_model.add(Dropout(0.3))\n# cnn_model.add(layers.Dense(32, activation='relu'))\n# cnn_model.add(BatchNormalization())\n# cnn_model.add(Dropout(0.3))\ncnn_model.add(layers.Dense(2, activation ='softmax'))\ncnn_model.summary()","metadata":{"id":"S92wdti_cHH6","outputId":"3fc54a1f-2fb0-4309-d9a2-8fdbbc8b5bc2","execution":{"iopub.status.busy":"2022-08-15T00:37:14.829757Z","iopub.execute_input":"2022-08-15T00:37:14.830274Z","iopub.status.idle":"2022-08-15T00:37:14.920482Z","shell.execute_reply.started":"2022-08-15T00:37:14.830226Z","shell.execute_reply":"2022-08-15T00:37:14.919081Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"cnn_model.compile(optimizer='adam',loss='categorical_crossentropy',metrics=['accuracy'],)","metadata":{"id":"EMoCr59t42Td","execution":{"iopub.status.busy":"2022-08-15T00:37:14.921912Z","iopub.execute_input":"2022-08-15T00:37:14.922614Z","iopub.status.idle":"2022-08-15T00:37:14.938957Z","shell.execute_reply.started":"2022-08-15T00:37:14.922575Z","shell.execute_reply":"2022-08-15T00:37:14.93771Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":" **1-Defining Callbacks**\n\n* A callback is an object that can perform actions at various stages of training (e.g. at the start or end of an epoch, before or after a single batch, etc)\n\n\n**2-Reduce Learning Rate on Plateau**\n* Is used to reduce the learning rate when a metric has stopped improving.\n\n","metadata":{"id":"ubaCfhsU_6QF"}},{"cell_type":"code","source":"from tensorflow.keras.callbacks import EarlyStopping,ReduceLROnPlateau\nearly = EarlyStopping(monitor=\"loss\", mode=\"min\",min_delta = 0,\n patience = 10,\n verbose = 1,\n restore_best_weights = True)\nlearning_rate_reduction = ReduceLROnPlateau(monitor='loss', patience = 2, verbose=1,factor=0.3, min_lr=0.000001)\ncallbacks_list = [ early, learning_rate_reduction]","metadata":{"id":"uxLfTTyhFnFy","execution":{"iopub.status.busy":"2022-08-15T00:37:14.940636Z","iopub.execute_input":"2022-08-15T00:37:14.941014Z","iopub.status.idle":"2022-08-15T00:37:14.956418Z","shell.execute_reply.started":"2022-08-15T00:37:14.94096Z","shell.execute_reply":"2022-08-15T00:37:14.954278Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Training model\n\nn_training_samples = len(train)\nn_validation_samples = len(validate)\nhistory = cnn_model.fit(\n train,\n epochs=60,\n validation_data=validate,\n validation_steps=n_validation_samples//batch_size,\n # steps_per_epoch =n_training_samples//batch_size,\n shuffle = True,\n callbacks=callbacks_list,\n class_weight=class_weights\n )","metadata":{"id":"CKSuE6FQ42WE","outputId":"a1c6e07b-52aa-4dce-df90-f21c8f199c9b","execution":{"iopub.status.busy":"2022-08-15T00:37:14.959076Z","iopub.execute_input":"2022-08-15T00:37:14.959662Z","iopub.status.idle":"2022-08-15T01:04:03.241595Z","shell.execute_reply.started":"2022-08-15T00:37:14.959609Z","shell.execute_reply":"2022-08-15T01:04:03.240302Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"score, acc = cnn_model.evaluate(validate,batch_size=batch_size) \nprint('Test score:', score)\nprint('Test accuracy:', acc)","metadata":{"id":"iqieAIee42Yy","outputId":"77175719-3974-4005-dd36-95caa0fcf615","execution":{"iopub.status.busy":"2022-08-15T01:04:03.24381Z","iopub.execute_input":"2022-08-15T01:04:03.244681Z","iopub.status.idle":"2022-08-15T01:04:05.999552Z","shell.execute_reply.started":"2022-08-15T01:04:03.24457Z","shell.execute_reply":"2022-08-15T01:04:05.998254Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print('Training Accuracy: {:.2f}%'.format(history.history[\"accuracy\"][-1:][0]*100))\nprint('Validation Accuracy: {:.2f}%'.format(history.history[\"val_accuracy\"][-1:][0]*100))\n","metadata":{"id":"nm4lluDAqsNY","execution":{"iopub.status.busy":"2022-08-15T01:04:06.001635Z","iopub.execute_input":"2022-08-15T01:04:06.002149Z","iopub.status.idle":"2022-08-15T01:04:06.009264Z","shell.execute_reply.started":"2022-08-15T01:04:06.002092Z","shell.execute_reply":"2022-08-15T01:04:06.007976Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print('Training Loss: {:.3f}'.format(history.history[\"loss\"][-1:][0]))\nprint('Validation Loss: {:.3f}'.format(history.history[\"val_loss\"][-1:][0]))\n","metadata":{"id":"qU8lHSMmRhu1","execution":{"iopub.status.busy":"2022-08-15T01:04:06.010728Z","iopub.execute_input":"2022-08-15T01:04:06.011978Z","iopub.status.idle":"2022-08-15T01:04:06.021917Z","shell.execute_reply.started":"2022-08-15T01:04:06.011928Z","shell.execute_reply":"2022-08-15T01:04:06.020669Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"cnn_prediction = cnn_model.predict(validate)\ncnn_prediction[0:5]","metadata":{"id":"yroCkc8e42eQ","outputId":"4bd20a96-f915-4cc2-c667-6c55a7a49c8c","execution":{"iopub.status.busy":"2022-08-15T01:04:06.023155Z","iopub.execute_input":"2022-08-15T01:04:06.023597Z","iopub.status.idle":"2022-08-15T01:04:08.824275Z","shell.execute_reply.started":"2022-08-15T01:04:06.023552Z","shell.execute_reply":"2022-08-15T01:04:08.823392Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"plt.plot(history.history['accuracy'], label='accuracy')\nplt.plot(history.history['val_accuracy'], label = 'val_accuracy')\nplt.xlabel('Epoch')\nplt.ylabel('Accuracy')\nplt.ylim([0.5, 1])\nplt.legend(loc='lower right')\nplt.show()","metadata":{"id":"5gWhTl-3PJK0","outputId":"c2d59fca-8942-4b6e-abad-25a41917abd7","execution":{"iopub.status.busy":"2022-08-15T01:04:08.825999Z","iopub.execute_input":"2022-08-15T01:04:08.827998Z","iopub.status.idle":"2022-08-15T01:04:09.03473Z","shell.execute_reply.started":"2022-08-15T01:04:08.827887Z","shell.execute_reply":"2022-08-15T01:04:09.033289Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"plt.plot(history.history[\"loss\"], label=\"train_loss\")\nplt.plot(history.history[\"val_loss\"], label=\"val_loss\")\nplt.title('Loss Plot')\nplt.xlabel(\"epochs\")\nplt.ylabel(\"loss\")\nplt.ylim([0.5, 1])\nplt.legend(loc='lower right')\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2022-08-15T01:04:09.036571Z","iopub.execute_input":"2022-08-15T01:04:09.037241Z","iopub.status.idle":"2022-08-15T01:04:09.23079Z","shell.execute_reply.started":"2022-08-15T01:04:09.037192Z","shell.execute_reply":"2022-08-15T01:04:09.229637Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"test_df ","metadata":{"id":"pU24REUsWsot","outputId":"43292598-c1ff-4f98-80b1-196f24442438","execution":{"iopub.status.busy":"2022-08-15T01:04:09.23245Z","iopub.execute_input":"2022-08-15T01:04:09.233131Z","iopub.status.idle":"2022-08-15T01:04:09.248632Z","shell.execute_reply.started":"2022-08-15T01:04:09.233084Z","shell.execute_reply":"2022-08-15T01:04:09.247408Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Predction Using CNN.","metadata":{"id":"LE7e42i4Wi6l"}},{"cell_type":"code","source":"test_gen = ImageDataGenerator(\n rescale = 1./255,\n \n )","metadata":{"id":"bM9aQDSKUdKJ","outputId":"55525d69-6679-404b-9cda-b7d558c0c0b0","execution":{"iopub.status.busy":"2022-08-15T01:04:09.250139Z","iopub.execute_input":"2022-08-15T01:04:09.250854Z","iopub.status.idle":"2022-08-15T01:04:09.259589Z","shell.execute_reply.started":"2022-08-15T01:04:09.250819Z","shell.execute_reply":"2022-08-15T01:04:09.258215Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"test = test_gen.flow_from_dataframe(\n test_df,\n # train_path_new+'/',\n test_path_new,\n x_col='new_image_name',\n # y_col =None,\n target_size=(96,96),\n class_mode=None, \n batch_size=batch_size,\n color_mode='rgb'\n )\n\n","metadata":{"id":"KzCxJToEPcnq","outputId":"cbd66454-108a-4fb8-cebe-d552faa9e079","execution":{"iopub.status.busy":"2022-08-15T01:04:09.261493Z","iopub.execute_input":"2022-08-15T01:04:09.26186Z","iopub.status.idle":"2022-08-15T01:04:09.276954Z","shell.execute_reply.started":"2022-08-15T01:04:09.261828Z","shell.execute_reply":"2022-08-15T01:04:09.275705Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"predection_test = cnn_model.predict(test)\npredection_test","metadata":{"id":"0IxBfBZ-MPp3","outputId":"0d9102ed-6a30-47c1-a0b8-514b08e273a5","execution":{"iopub.status.busy":"2022-08-15T01:04:09.27912Z","iopub.execute_input":"2022-08-15T01:04:09.279578Z","iopub.status.idle":"2022-08-15T01:04:09.570062Z","shell.execute_reply.started":"2022-08-15T01:04:09.279544Z","shell.execute_reply":"2022-08-15T01:04:09.568841Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"pred= np.argmax(predection_test,axis=1)\npred","metadata":{"execution":{"iopub.status.busy":"2022-08-15T01:04:09.571861Z","iopub.execute_input":"2022-08-15T01:04:09.572358Z","iopub.status.idle":"2022-08-15T01:04:09.580471Z","shell.execute_reply.started":"2022-08-15T01:04:09.572309Z","shell.execute_reply":"2022-08-15T01:04:09.579406Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"sub_df = pd.DataFrame(predection_test, columns=[\"CE\", \"LAA\"])\nsub_df.head()","metadata":{"execution":{"iopub.status.busy":"2022-08-15T01:04:09.581741Z","iopub.execute_input":"2022-08-15T01:04:09.58208Z","iopub.status.idle":"2022-08-15T01:04:09.599375Z","shell.execute_reply.started":"2022-08-15T01:04:09.58205Z","shell.execute_reply":"2022-08-15T01:04:09.598089Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"test_filenames = test.filenames\ntest_filenames","metadata":{"execution":{"iopub.status.busy":"2022-08-15T01:04:09.601019Z","iopub.execute_input":"2022-08-15T01:04:09.60147Z","iopub.status.idle":"2022-08-15T01:04:09.612339Z","shell.execute_reply.started":"2022-08-15T01:04:09.601426Z","shell.execute_reply":"2022-08-15T01:04:09.611064Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"sub_df[\"image_name\"] = test_filenames\nsub_df","metadata":{"execution":{"iopub.status.busy":"2022-08-15T01:04:09.613991Z","iopub.execute_input":"2022-08-15T01:04:09.614588Z","iopub.status.idle":"2022-08-15T01:04:09.63204Z","shell.execute_reply.started":"2022-08-15T01:04:09.614553Z","shell.execute_reply":"2022-08-15T01:04:09.631098Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"sub_df[\"patient_id\"] = sub_df[\"image_name\"].apply(get_id)\nsub_df","metadata":{"execution":{"iopub.status.busy":"2022-08-15T01:04:09.634779Z","iopub.execute_input":"2022-08-15T01:04:09.635161Z","iopub.status.idle":"2022-08-15T01:04:09.65002Z","shell.execute_reply.started":"2022-08-15T01:04:09.635127Z","shell.execute_reply":"2022-08-15T01:04:09.648739Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"sub_df = sub_df.reindex(columns=['patient_id', 'CE', 'LAA'])\nsub_df","metadata":{"execution":{"iopub.status.busy":"2022-08-15T01:04:09.651949Z","iopub.execute_input":"2022-08-15T01:04:09.652418Z","iopub.status.idle":"2022-08-15T01:04:09.667489Z","shell.execute_reply.started":"2022-08-15T01:04:09.652372Z","shell.execute_reply":"2022-08-15T01:04:09.666546Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"sub_df = sub_df.reindex(columns=['patient_id', 'CE', 'LAA'])\nsub_df","metadata":{"execution":{"iopub.status.busy":"2022-08-15T01:04:09.668816Z","iopub.execute_input":"2022-08-15T01:04:09.669722Z","iopub.status.idle":"2022-08-15T01:04:09.685385Z","shell.execute_reply.started":"2022-08-15T01:04:09.669675Z","shell.execute_reply":"2022-08-15T01:04:09.683932Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"sub_df = sub_df.groupby(\"patient_id\").mean()\nsub_df[[\"CE\", \"LAA\"]].round(6).to_csv(\"submission.csv\")\n!head submission.csv","metadata":{"execution":{"iopub.status.busy":"2022-08-15T01:04:09.686944Z","iopub.execute_input":"2022-08-15T01:04:09.68835Z","iopub.status.idle":"2022-08-15T01:04:10.936918Z","shell.execute_reply.started":"2022-08-15T01:04:09.688311Z","shell.execute_reply":"2022-08-15T01:04:10.935527Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# sub_df[[\"patient_id\", \"CE\", \"LAA\"]].to_csv(\"submission.csv\", index=False)\n# !head submission.csv","metadata":{"execution":{"iopub.status.busy":"2022-08-15T01:04:10.938974Z","iopub.execute_input":"2022-08-15T01:04:10.939375Z","iopub.status.idle":"2022-08-15T01:04:10.945241Z","shell.execute_reply.started":"2022-08-15T01:04:10.93934Z","shell.execute_reply":"2022-08-15T01:04:10.943948Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from IPython.display import FileLink\nFileLink(r'submission.csv')","metadata":{"execution":{"iopub.status.busy":"2022-08-15T01:04:10.946826Z","iopub.execute_input":"2022-08-15T01:04:10.947425Z","iopub.status.idle":"2022-08-15T01:04:10.96113Z","shell.execute_reply.started":"2022-08-15T01:04:10.947385Z","shell.execute_reply":"2022-08-15T01:04:10.959529Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}