Image Source: ProjectGurukul<\/a><\/em><\/p>\n So our main aim in this project is that with the help of a dataset we will create a model which will correctly classify whether the Breast Cancer is of malignant or benign type.<\/p>\n We will be using a breast cancer dataset which you can download\u00a0 from this link: Breast Cancer Dataset<\/strong><\/a><\/p>\n We are going to analyze the dataset completely, which will clear all your questions regarding what dataset we will be using, how many rows and columns are there, etc.<\/p>\n So there are 10 columns that are:<\/p>\n I have worked on google collabs, if you work on your system please install the following libraries:<\/p>\n To install, open your command prompt and run:<\/p>\n Please download the source code of breast cancer classification using machine learning: Breast Cancer Classification Project Code<\/strong><\/a><\/p>\n Firstly we have to import all the required libraries that we have installed above. We will also import some libraries at the time we will use them.<\/p>\n If you are using google collabs you have to first upload the dataset to access that data. So to upload the dataset run following command:<\/p>\n If you are using a jupyter notebook or working on your system, just read our dataset using read_csv() method.<\/p>\n Now we will analyse our dataset to see what is the shape of our data, how many empty values are present in our dataset, and we will drop those missing values using various methods provided by pandas.<\/p>\n Let\u2019s create a pairplot that will show us the complete relationship between radius mean, texture mean, perimeter mean, area mean and smoothness mean on the basis of diagnosis type.<\/strong><\/p>\n Now we will visualize the diagnosis column in our dataset to see how many malignant and benign are present.<\/strong><\/p>\n In this whole analyzing process we are going to convert our data and perform some data processing so that we can build a model which can classify the type of Breast cancer using this preprocessed data.<\/p>\n I have written comments above each line of code about what we are doing and why we are doing so that you can understand it better and easily.<\/p>\n Now we will convert our text (B and M) to integers (0 and 1) using LabelEncoder provided by sklearn library.<\/p>\n X will be our main features data which consists of all the columns except the label and id column. We also normalize our X data using MinMaxScaler provided by sklearn library.<\/p>\n Now we will split our data into training data and testing data using train_test_split.<\/p>\n In this step, we will be creating a Sequential model with the help of TensorFlow and Keras. In the model we have created three Dense layers in which one is the input layer with 128 hidden layers and the activation function is relu, and the second layer consists of 64 hidden layers with activation function relu and the third layer is final output layer and activation function is sigmoid.<\/p>\n Compile the model and view a summary of the model<\/p>\n Fit the model to see the accuracy of training data:<\/strong><\/p>\n In this step, we will be analyzing our training accuracy and validation accuracy and also we will be plotting losses at each epoch with the help of matplotlib library.<\/p>\n![\"breast](\"https:\/\/techvidvan.com\/tutorials\/wp-content\/uploads\/sites\/2\/2021\/08\/breast-cancer-diagram.jpg\")
<\/a><\/p>\nBreast Cancer Dataset<\/h3>\n
\n
Project Prerequisites<\/h3>\n
\n
pip install numpy\r\npip install tensorflow\r\npip install pandas\r\npip install sklearn<\/pre>\n
Download Breast Cancer Classification Project Code<\/h3>\n
Now let’s start Analysing and Implementing our Breast Cancer Classification Project by TechVidvan:<\/h3>\n
1) Importing Libraries:<\/h4>\n
# Import libraries for Breast Cancer Classification Project\r\nimport numpy as np\r\nimport pandas as pd\r\nimport matplotlib.pyplot as plt\r\nimport seaborn as sns\r\n<\/pre>\n
2) Loading the dataset:<\/h4>\n
# Load the dataset\r\nfrom google.colab import files\r\nuploaded = files.upload()\r\n<\/pre>\n
<\/a><\/p>\ndf = pd.read_csv('data.csv')\r\n\r\n#Now let\u2019s view our dataset using head():\r\n\r\ndf.head(10)\r\n<\/pre>\n<\/a><\/p>\n3) Analysing the data:<\/h4>\n
# count the number of rows and columns in dataset:\r\ndf.shape\r\n<\/pre>\n
sns.pairplot(df,hue = 'diagnosis', palette= 'coolwarm', vars = ['radius_mean', 'texture_mean', 'perimeter_mean','area_mean','smoothness_mean'])\r\n<\/pre>\n
<\/a><\/p>\n# count the number of empty values in each columns:\r\ndf.isna().sum()\r\n\r\n# drop the columns with all the missing values:\r\ndf = df.dropna(axis = 1)\r\n\r\ndf.shape\r\n\r\n# Get the count of the number of Malignant(M) or Benign(B) cells\r\ndf['diagnosis'].value_counts()\r\n<\/pre>\n
# visualize the count:\r\nsns.countplot(df['diagnosis'], label = 'count')\r\n<\/pre>\n
<\/a><\/p>\n# look at the data types to see which columns need to be encoded:\r\ndf.dtypes\r\n<\/pre>\n
<\/a><\/p>\n# Rename the diagnosis data to labels:\r\ndf = df.rename(columns = {'diagnosis' : 'label'})\r\nprint(df.dtypes)\r\n\r\n<\/pre>\n<\/a><\/p>\n# define the dependent variable that need to predict(label)\r\ny = df['label'].values\r\nprint(np.unique(y))\r\n<\/pre>\n
4) Encoding Categorical Data:<\/h4>\n
# Encoding categorical data from text(B and M) to integers (0 and 1)\r\nfrom sklearn.preprocessing import LabelEncoder\r\nlabelencoder = LabelEncoder()\r\nY = labelencoder.fit_transform(y) # M = 1 and B = 0\r\nprint(np.unique(Y))\r\n<\/pre>\n
5) Defining X :<\/h4>\n
# define x and normalize \/ scale value:\r\n\r\n# define the independent variables, Drop label and ID, and normalize other data:\r\nX = df.drop(labels=['label','id'],axis = 1)\r\n\r\n#scale \/ normalize the values to bring them into similar range:\r\nfrom sklearn.preprocessing import MinMaxScaler\r\nscaler = MinMaxScaler()\r\nscaler.fit(X)\r\nX = scaler.transform(X)\r\n\r\nprint(X)\r\n<\/pre>\n
6) Splitting Our data:<\/h4>\n
# Split data into training and testing data to verify accuracy after fitting the model\r\nfrom sklearn.model_selection import train_test_split\r\nx_train,x_test,y_train,y_test = train_test_split(X,Y, test_size = 0.25, random_state=42)\r\nprint('Shape of training data is: ', x_train.shape)\r\nprint('Shape of testing data is: ', x_test.shape)\r\n<\/pre>\n7) Creating Model:<\/h4>\n
import tensorflow\r\nfrom tensorflow.keras.models import Sequential\r\nfrom tensorflow.keras.layers import Dense, Activation, Dropout\r\n\r\nmodel = Sequential()\r\nmodel.add(Dense(128, input_dim=30, activation='relu'))\r\nmodel.add(Dropout(0.5))\r\nmodel.add(Dense(64,activation = 'relu'))\r\nmodel.add(Dropout(0.5))\r\nmodel.add(Dense(1))\r\nmodel.add(Activation('sigmoid'))\r\n<\/pre>\n8) Compile and fit ml model to our training data:<\/h4>\n
model.compile(loss = 'binary_crossentropy', optimizer = 'adam' , metrics = ['accuracy'])\r\n\r\nmodel.summary()\r\n<\/pre>\n
<\/a><\/p>\n# fit with no early stopping or other callbacks:\r\nhistory = model.fit(x_train,y_train,verbose = 1,epochs = 100, batch_size = 64,validation_data = (x_test,y_test))\r\n<\/pre>\n
<\/a><\/p>\n9) Visualizing our training accuracy and validation accuracy:<\/h4>\n
# plot the training and validation accuracy and loss at each epochs:\r\nloss = history.history['loss']\r\nval_loss = history.history['val_loss']\r\nepochs = range(1,len(loss)+1)\r\nplt.plot(epochs,loss,'y',label = 'Training loss')\r\nplt.plot(epochs,val_loss,'r',label = 'Validation loss')\r\nplt.title('TechVidvan Training and Validation loss')\r\nplt.xlabel('Epochs')\r\nplt.ylabel('Loss')\r\nplt.legend()\r\nplt.show()\r\n\r\nacc = history.history['accuracy']\r\nval_acc = history.history['val_accuracy']\r\nplt.plot(epochs,acc,'y',label = 'Training acc')\r\nplt.plot(epochs,val_acc,'r',label = 'Validation acc')\r\nplt.title('TechVidvan Training and Validation accuracy')\r\nplt.xlabel('Epochs')\r\nplt.ylabel('Accuracy')\r\nplt.legend()\r\nplt.show()\r\n<\/pre>\n