brian-kipkoech-tanui / expenditurechurnannbinaryclassification

Artificial Neural Network Classification

ANN Classification

Importing the libraries

import numpy as np
import pandas as pd
import tensorflow as tf

tf.__version__

'2.11.0'

Data Preprocessing

Importing the Dataset

dataset = pd.read_excel("./Expenditure-churn (3).xlsx")
dataset.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 10000 entries, 0 to 9999
Data columns (total 12 columns):
 #   Column    Non-Null Count  Dtype  
---  ------    --------------  -----  
 0   age       10000 non-null  int64  
 1   gender    10000 non-null  int64  
 2   marital   10000 non-null  float64
 3   dep       10000 non-null  int64  
 4   Income    10000 non-null  float64
 5   Job yrs   10000 non-null  int64  
 6   Town yrs  10000 non-null  int64  
 7   Yrs Ed    10000 non-null  int64  
 8   Dri Lic   10000 non-null  int64  
 9   Own Home  10000 non-null  int64  
 10  # Cred C  10000 non-null  int64  
 11  Churn     10000 non-null  int64  
dtypes: float64(2), int64(10)
memory usage: 937.6 KB

X = dataset.iloc[:,:-1].values
y = dataset.iloc[:,-1].values

Splitting the data into Training and Test set

from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.25, random_state=0)

Feature Scaling

from sklearn.preprocessing import StandardScaler
sc = StandardScaler()
X_train = sc.fit_transform(X_train)
X_test = sc.transform(X_test)

Building the ANN

Initializing the ann

ann = tf.keras.models.Sequential()

Adding the input layer and the first hidden layer

ann.add(tf.keras.layers.Dense(units=64, activation='relu')) 

Adding the second hidden layer

ann.add(tf.keras.layers.Dense(units=48, activation='relu'))

Adding the Output layer

ann.add(tf.keras.layers.Dense(units=1, activation='sigmoid'))

Training the ANN

Compiling the ANN

ann.compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = ['accuracy'])

Training the ANN on the Training Set

ann.fit(X_train, y_train, batch_size=32, epochs=100)

Epoch 1/100
235/235 [==============================] - 3s 4ms/step - loss: 0.2616 - accuracy: 0.8872
Epoch 2/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0635 - accuracy: 0.9825
Epoch 3/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0398 - accuracy: 0.9871
Epoch 4/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0300 - accuracy: 0.9904
Epoch 5/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0234 - accuracy: 0.9927
Epoch 6/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0192 - accuracy: 0.9943
Epoch 7/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0166 - accuracy: 0.9956
Epoch 8/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0144 - accuracy: 0.9959
Epoch 9/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0126 - accuracy: 0.9968
Epoch 10/100
235/235 [==============================] - 1s 4ms/step - loss: 0.0114 - accuracy: 0.9961
Epoch 11/100
235/235 [==============================] - 1s 4ms/step - loss: 0.0106 - accuracy: 0.9968
Epoch 12/100
235/235 [==============================] - 1s 4ms/step - loss: 0.0087 - accuracy: 0.9977
Epoch 13/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0102 - accuracy: 0.9965
Epoch 14/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0075 - accuracy: 0.9977
Epoch 15/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0064 - accuracy: 0.9980
Epoch 16/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0067 - accuracy: 0.9984
Epoch 17/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0053 - accuracy: 0.9988
Epoch 18/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0045 - accuracy: 0.9991
Epoch 19/100
235/235 [==============================] - 1s 4ms/step - loss: 0.0041 - accuracy: 0.9989
Epoch 20/100
235/235 [==============================] - 1s 4ms/step - loss: 0.0065 - accuracy: 0.9975
Epoch 21/100
235/235 [==============================] - 1s 4ms/step - loss: 0.0048 - accuracy: 0.9987
Epoch 22/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0038 - accuracy: 0.9988
Epoch 23/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0026 - accuracy: 0.9995
Epoch 24/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0043 - accuracy: 0.9984
Epoch 25/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0020 - accuracy: 0.9999
Epoch 26/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0023 - accuracy: 0.9996
Epoch 27/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0032 - accuracy: 0.9989
Epoch 28/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0048 - accuracy: 0.9984
Epoch 29/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0050 - accuracy: 0.9981
Epoch 30/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0018 - accuracy: 0.9997
Epoch 31/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0018 - accuracy: 0.9996
Epoch 32/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0018 - accuracy: 0.9997
Epoch 33/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0016 - accuracy: 0.9999
Epoch 34/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0022 - accuracy: 0.9996
Epoch 35/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0023 - accuracy: 0.9993
Epoch 36/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0069 - accuracy: 0.9975
Epoch 37/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0043 - accuracy: 0.9981
Epoch 38/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0018 - accuracy: 0.9996
Epoch 39/100
235/235 [==============================] - 1s 3ms/step - loss: 8.0494e-04 - accuracy: 0.9999
Epoch 40/100
235/235 [==============================] - 1s 3ms/step - loss: 5.9393e-04 - accuracy: 1.0000
Epoch 41/100
235/235 [==============================] - 1s 3ms/step - loss: 5.7110e-04 - accuracy: 1.0000
Epoch 42/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0029 - accuracy: 0.9987
Epoch 43/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0028 - accuracy: 0.9992
Epoch 44/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0015 - accuracy: 0.9995
Epoch 45/100
235/235 [==============================] - 1s 3ms/step - loss: 4.0946e-04 - accuracy: 1.0000
Epoch 46/100
235/235 [==============================] - 1s 3ms/step - loss: 4.3892e-04 - accuracy: 1.0000
Epoch 47/100
235/235 [==============================] - 1s 3ms/step - loss: 3.9667e-04 - accuracy: 1.0000
Epoch 48/100
235/235 [==============================] - 1s 3ms/step - loss: 3.2670e-04 - accuracy: 1.0000
Epoch 49/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0066 - accuracy: 0.9981
Epoch 50/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0062 - accuracy: 0.9981
Epoch 51/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0040 - accuracy: 0.9988
Epoch 52/100
235/235 [==============================] - 1s 3ms/step - loss: 4.7653e-04 - accuracy: 1.0000
Epoch 53/100
235/235 [==============================] - 1s 3ms/step - loss: 3.1398e-04 - accuracy: 1.0000
Epoch 54/100
235/235 [==============================] - 1s 3ms/step - loss: 2.4859e-04 - accuracy: 1.0000
Epoch 55/100
235/235 [==============================] - 1s 3ms/step - loss: 3.7687e-04 - accuracy: 1.0000
Epoch 56/100
235/235 [==============================] - 1s 3ms/step - loss: 2.8085e-04 - accuracy: 1.0000
Epoch 57/100
235/235 [==============================] - 1s 3ms/step - loss: 1.9511e-04 - accuracy: 1.0000
Epoch 58/100
235/235 [==============================] - 1s 3ms/step - loss: 1.7072e-04 - accuracy: 1.0000
Epoch 59/100
235/235 [==============================] - 1s 3ms/step - loss: 1.6152e-04 - accuracy: 1.0000
Epoch 60/100
235/235 [==============================] - 1s 3ms/step - loss: 1.6068e-04 - accuracy: 1.0000
Epoch 61/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0107 - accuracy: 0.9973
Epoch 62/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0038 - accuracy: 0.9981
Epoch 63/100
235/235 [==============================] - 1s 3ms/step - loss: 3.4746e-04 - accuracy: 1.0000
Epoch 64/100
235/235 [==============================] - 1s 3ms/step - loss: 2.5776e-04 - accuracy: 1.0000
Epoch 65/100
235/235 [==============================] - 1s 3ms/step - loss: 2.0422e-04 - accuracy: 1.0000
Epoch 66/100
235/235 [==============================] - 1s 3ms/step - loss: 1.5546e-04 - accuracy: 1.0000
Epoch 67/100
235/235 [==============================] - 1s 3ms/step - loss: 1.5003e-04 - accuracy: 1.0000
Epoch 68/100
235/235 [==============================] - 1s 3ms/step - loss: 1.3928e-04 - accuracy: 1.0000
Epoch 69/100
235/235 [==============================] - 1s 3ms/step - loss: 1.3602e-04 - accuracy: 1.0000
Epoch 70/100
235/235 [==============================] - 1s 3ms/step - loss: 1.8484e-04 - accuracy: 1.0000
Epoch 71/100
235/235 [==============================] - 1s 3ms/step - loss: 4.6412e-04 - accuracy: 0.9999
Epoch 72/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0033 - accuracy: 0.9987
Epoch 73/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0093 - accuracy: 0.9976
Epoch 74/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0022 - accuracy: 0.9993
Epoch 75/100
235/235 [==============================] - 1s 3ms/step - loss: 3.5701e-04 - accuracy: 1.0000
Epoch 76/100
235/235 [==============================] - 1s 3ms/step - loss: 1.2967e-04 - accuracy: 1.0000
Epoch 77/100
235/235 [==============================] - 1s 3ms/step - loss: 9.9854e-05 - accuracy: 1.0000
Epoch 78/100
235/235 [==============================] - 1s 3ms/step - loss: 9.9170e-05 - accuracy: 1.0000
Epoch 79/100
235/235 [==============================] - 1s 3ms/step - loss: 8.3809e-05 - accuracy: 1.0000
Epoch 80/100
235/235 [==============================] - 1s 3ms/step - loss: 9.6040e-05 - accuracy: 1.0000
Epoch 81/100
235/235 [==============================] - 1s 3ms/step - loss: 1.0550e-04 - accuracy: 1.0000
Epoch 82/100
235/235 [==============================] - 1s 3ms/step - loss: 1.0744e-04 - accuracy: 1.0000
Epoch 83/100
235/235 [==============================] - 1s 3ms/step - loss: 1.3892e-04 - accuracy: 1.0000
Epoch 84/100
235/235 [==============================] - 1s 3ms/step - loss: 1.0314e-04 - accuracy: 1.0000
Epoch 85/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0160 - accuracy: 0.9975
Epoch 86/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0038 - accuracy: 0.9985
Epoch 87/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0020 - accuracy: 0.9991
Epoch 88/100
235/235 [==============================] - 1s 3ms/step - loss: 0.0037 - accuracy: 0.9987
Epoch 89/100
235/235 [==============================] - 1s 3ms/step - loss: 4.8882e-04 - accuracy: 1.0000
Epoch 90/100
235/235 [==============================] - 1s 3ms/step - loss: 2.2288e-04 - accuracy: 1.0000
Epoch 91/100
235/235 [==============================] - 1s 3ms/step - loss: 1.1115e-04 - accuracy: 1.0000
Epoch 92/100
235/235 [==============================] - 1s 3ms/step - loss: 9.5189e-05 - accuracy: 1.0000
Epoch 93/100
235/235 [==============================] - 1s 3ms/step - loss: 1.0408e-04 - accuracy: 1.0000
Epoch 94/100
235/235 [==============================] - 1s 3ms/step - loss: 8.0245e-05 - accuracy: 1.0000
Epoch 95/100
235/235 [==============================] - 1s 3ms/step - loss: 9.1883e-05 - accuracy: 1.0000
Epoch 96/100
235/235 [==============================] - 1s 3ms/step - loss: 7.3396e-05 - accuracy: 1.0000
Epoch 97/100
235/235 [==============================] - 1s 3ms/step - loss: 9.4839e-05 - accuracy: 1.0000
Epoch 98/100
235/235 [==============================] - 1s 3ms/step - loss: 7.1007e-05 - accuracy: 1.0000
Epoch 99/100
235/235 [==============================] - 1s 3ms/step - loss: 8.0511e-05 - accuracy: 1.0000
Epoch 100/100
235/235 [==============================] - 1s 3ms/step - loss: 6.7545e-05 - accuracy: 1.0000





<keras.callbacks.History at 0x1b04584f788>

Making Predictions and evaluating the model

Predicting the test set results

y_pred = ann.predict(X_test)
y_pred = (y_pred>0.5)
print(np.concatenate((y_pred.reshape(len(y_pred),1), y_test.reshape(len(y_test),1)),1))

79/79 [==============================] - 0s 2ms/step
[[0 0]
 [0 0]
 [0 0]
 ...
 [0 0]
 [0 0]
 [0 0]]

Making the Confusion Matrix

from sklearn.metrics import confusion_matrix, accuracy_score, precision_score, recall_score, roc_auc_score
cm = confusion_matrix(y_test, y_pred)
# Print the errors
print("Accuracy Score:   "+str(accuracy_score(y_pred, y_test)*100))
print("Precision Score:  "+str(precision_score(y_pred, y_test)*100))
print("Recall Score:     "+str(recall_score(y_pred, y_test)*100))
print("roc_auc_score:    "+str(accuracy_score(y_pred, y_test)*100))
print("\nConfusion Matrix:\n", confusion_matrix(y_pred, y_test))

Accuracy Score:   99.6
Precision Score:  99.26650366748166
Recall Score:     99.50980392156863
roc_auc_score:    99.6

Confusion Matrix:
 [[1678    6]
 [   4  812]]