ANN explained with Realtime example

Introduction:

The ANN neural network contains three layers, The first one is input layer and the last one is the output layer, IN between the input and output layer there will be two hidden layers. You can ask one question like, What happens if I add extra layers, I have also the same doubt while making this script, The Information about the layers are,You can add multiple numbers of hidden layers, But You can add only 1 input and 1 output layer, You can’t add more number input and output layers. So it is the general structure of ANN, Next i will explain you what are things running behind the each and every layer.

First layer of structure is input layer, As the name suggests, It only receive the different format inputs from the programmer, It is basic duty of input layer in every deep learning neural network, And the second layers are the hidden layers, The name itself says like it is hidden layer, It will find the hidden features and patterns behind every input. Final one is output layer, It will collect the transformations of input data that operated in hidden layer and it present the data results to user. This is the general workflow of the ANN,

Source code:

Imports:

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from sklearn import preprocessing
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split
import seaborn as sns
from keras.layers import Dense, BatchNormalization, Dropout, LSTM
from keras.models import Sequential
from tensorflow.keras.utils import to_categorical
from keras import callbacks
from sklearn.metrics import precision_score, recall_score, confusion_matrix, classification_report, accuracy_score, f1_score

Load data:

#loading data
data = pd.read_csv("/content/heart_failure_clinical_records_dataset.csv")
data.head()

Feature selection:
#assigning values to features as X and target as y
X=data.drop(["DEATH_EVENT"],axis=1)
y=data["DEATH_EVENT"]
Scaling:
#Set up a standard scaler for the features
col_names = list(X.columns)
s_scaler = preprocessing.StandardScaler()
X_df= s_scaler.fit_transform(X)
X_df = pd.DataFrame(X_df, columns=col_names)   
X_df.describe().T

Splitting data:
#spliting test and training sets
X_train, X_test, y_train,y_test = train_test_split(X_df,y,test_size=0.25,random_state=7)

Model Structure:
early_stopping = callbacks.EarlyStopping(
    min_delta=0.001, # minimium amount of change to count as an improvement
    patience=30, # how many epochs to wait before stopping
    restore_best_weights=True)

# Initialising the NN
model = Sequential()

# layers
model.add(Dense(units = 16, kernel_initializer = 'uniform', activation = 'relu', input_dim = 12))
model.add(Dense(units = 8, kernel_initializer = 'uniform', activation = 'relu'))
model.add(Dropout(0.25))
model.add(Dense(units = 4, kernel_initializer = 'uniform', activation = 'relu'))
model.add(Dropout(0.5))
model.add(Dense(units = 1, kernel_initializer = 'uniform', activation = 'sigmoid'))
from tensorflow.keras.optimizers import SGD
# Compiling the ANN
model.compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = ['accuracy'])

Training model:
history = model.fit(X_train, y_train, batch_size = 32, epochs = 500,callbacks=[early_stopping], validation_split=0.2)
Prediction:
# Predicting the test set results
y_pred = model.predict(X_test)
y_pred = (y_pred > 0.5)
np.set_printoptions()
Confusion matrix:
cmap1 = sns.diverging_palette(275,150,  s=40, l=65, n=6)
plt.subplots(figsize=(12,8))
cf_matrix = confusion_matrix(y_test, y_pred)
sns.heatmap(cf_matrix/np.sum(cf_matrix), cmap = cmap1, annot = True, annot_kws = {'size':15})

Thanks for reading article,Support..!