In [10]: df['Exited' ].

value_counts ()

Import Libraries Female 4544

Name: count, dtype: int64
In [3]: import numpy as np
import pandas as pd In [13]: df. drop(columns =['RowNumber' , 'CustomerId' ,'Surname' ],inplace =True )

In [4]: df=pd .read_csv ("Churn_Modelling.csv" ) In [15]: df = pd.get_dummies ( df, columns =['Geography','Gender' ], prefix=[ 'Geography' ,
df.head () 'Gender' ],drop_first = False , dtype = int)

RowNumber CustomerId Surname CreditScore Geography Gender Age Tenure Balance Nu In [16]: X = df.drop (columns= ['Exited' ], axis = 1)
y = df['Exited' ]
0 1 15634602 Hargrave 619 France Female 42.0 2 0.00
Out[4]:
1 2 15647311 Hill 608 Spain Female 41.0 1 83807.86 In [17]: from sklearn .model_selection import train_test_split
2 3 15619304 Onio 502 France Female 42.0 8 159660.80 X_train ,X_test ,y_train ,y_test = train_test_split (X,y,test_size =0.2 ,random_state =4

3 4 15701354 Boni 699 France Female 39.0 1 0.00

4 5 15737888 Mitchell 850 Spain Female 43.0 2 125510.82 In [30]: from sklearn .preprocessing import StandardScaler
scaler = StandardScaler ()
X_train = scaler .fit_transform (X_train )
X_test = scaler .transform (X_test )
In [5]: df.shape

Out[5]: (10002, 14)

In [6]: df.info ()

<class 'pandas.core.frame.DataFrame' >

RangeIndex: 10002 entries, 0 to 10001
Data columns (total 14 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 RowNumber 10002 non-null int64
1 CustomerId 10002 non-null int64
2 Surname 10002 non-null object
3 CreditScore 10002 non-null int64
4 Geography 10001 non-null object
5 Gender 10002 non-null object
6 Age 10001 non-null float64
7 Tenure 10002 non-null int64
8 Balance 10002 non-null float64
9 NumOfProducts 10002 non-null int64
10 HasCrCard 10001 non-null float64
11 IsActiveMember 10001 non-null float64
12 EstimatedSalary 10002 non-null float64
13 Exited 10002 non-null int64 dtypes: float64(5),
int64(6), object(3) memory usage: 1.1+ MB
In [38]: model .fit(X_train , y_train ,epochs =10)
In [7]: df.duplicated ().sum()

Epoch 1/10
Out[7]: 2
251/251 [==============================] - 0s 2ms/step - loss: nan - accuracy:
0.7955
Epoch 2/10
251/251 [==============================] - 0s 1ms/step - loss: nan - accuracy:
0.7955
Epoch 3/10
251/251 [==============================] - 0s 1ms/step - loss: nan - accuracy:
0.7955
Epoch 4/10
251/251 [==============================] - 0s 2ms/step - loss: nan - accuracy:
0.7955
Epoch 5/10
251/251 [==============================] - 0s 2ms/step - loss: nan - accuracy:
In [35]: import tensorflow from tensorflow import
0.7955
keras from tensorflow.keras import
Epoch 6/10
Sequential from tensorflow.keras.layers
251/251 [==============================] - 0s 1ms/step - loss: nan - accuracy:
import Dense model = Sequential()
0.7955
model.add(Dense(units=3, activation='sigmoid', input_dim = 13))
Epoch 7/10
model.add(Dense(units=1, activation='sigmoid')) model.summary()
251/251 [==============================] - 0s 1ms/step - loss: nan - accuracy:
Model: "sequential_3" 0.7955
_________________________________________________________________ Epoch 8/10
Layer (type) Output Shape Param # 251/251 [==============================] - 0s 2ms/step - loss: nan - accuracy:
================================================================= 0.7955
dense_6 (Dense) (None, 3) 42 Epoch 9/10
dense_7 (Dense) (None, 1) 4 251/251 [==============================] - 0s 2ms/step - loss: nan - accuracy:
================================================================= 0.7955
Total params: 46 (184.00 Byte) Epoch 10/10
Trainable params: 46 (184.00 Byte) 251/251 [==============================] - 0s 2ms/step - loss: nan - accuracy:
Non-trainable params: 0 (0.00 Byte) 0.7955
_________________________________________________________________
Out[38]: <keras.src.callbacks.History at 0x2cd0e2a99c0>

In [36]: from tensorflow.keras.optimizers import Adam In [40]: model .layers [1].get_weights ()

model.compile(optimizer=Adam(learning_rate=0.001),
loss='binary_crossentropy', Out[40]: [array([[nan],
metrics=['accuracy']) [nan],
Out[10]: Exited [nan]], dtype=float32),
0 7964 In [45]: print("classification Report", classification_report (y_test, y_pred))
1 2038
Name: count, dtype: int64 classification Report precision recall f1-score support

In [11]: df[ 'Geography' ].value_counts () 0 0.80 1.00 0.89 1599

1 0.00 0.00 0.00 402
Out[11]: Geography
France 5014 accuracy 0.80 2001
Germany 2510 macro avg 0.40 0.50 0.44 2001
Spain 2477 weighted avg 0.64 0.80 0.71 2001
Name: count, dtype: int64

In [12]: df[ 'Gender']. value_counts () C:\Users\Harleen Manmeet\AppData\Local\Programs\Python\Python310\lib\site-

packag es\sklearn\metrics\_classification.py:1344: UndefinedMetricWarning:
Precision an d F-score are ill-defined and being set to 0.0 in labels with no
Out[12]: Gender
predicted sampl es. Use `zero_division` parameter to control this behavior.
Male 5458
_warn_prf(average, modifier, msg_start, len(result))
array([nan], dtype=float32)]

In [41]: y_log = model .predict( X_test)

63/63 [==============================] - 0s 1ms/step

In [42]: y_pred = np.where (y_log > 0.5,1 ,0)

In [43]: from sklearn.metrics import accuracy_score , classification_report

In [44]: accuracy_score (y_test,y_pred )

Out[44]: 0.7991004497751124
 C:\Users\Harleen
Manmeet\AppData\Local\Programs\Python\Python310\lib\site-packag
es\sklearn\metrics\_classification.py:1344:
UndefinedMetricWarning: Precision an d F-score are ill-defined
and being set to 0.0 in labels with no predicted sampl es. Use
`zero_division` parameter to control this behavior.
_warn_prf(average, modifier, msg_start, len(result))
C:\Users\Harleen
Manmeet\AppData\Local\Programs\Python\Python310\lib\site-packag
es\sklearn\metrics\_classification.py:1344:
UndefinedMetricWarning: Precision an d F-score are ill-defined
and being set to 0.0 in labels with no predicted sampl es. Use
`zero_division` parameter to control this behavior.
_warn_prf(average, modifier, msg_start, len(result))

In [ ]: