Обнаружение фрод

import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score, confusion_matrix

# Read in the data from the CSV file
df = pd.read_csv('https://raw.githubusercontent.com/dm-fedorov/ml/master/datasets/payment_fraud.csv')

df.sample(3)

	accountAgeDays	numItems	localTime	paymentMethod	paymentMethodAgeDays	label
31814	4	1	4.965339	creditcard	3.322222	0
27866	5	1	5.017904	creditcard	4.065278	0
33957	506	2	4.745402	creditcard	0.000000	0

# Convert categorical feature into dummy variables with one-hot encoding
df = pd.get_dummies(df, columns=['paymentMethod'])
df.sample(3)

	accountAgeDays	numItems	localTime	paymentMethodAgeDays	label	paymentMethod_creditcard	paymentMethod_paypal	paymentMethod_storecredit
25716	975	1	5.040929	0.000000	0	1	0	0
36918	19	1	4.057414	0.001389	0	1	0	0
29477	1001	1	4.505662	0.000000	0	1	0	0

# Split dataset up into train and test sets
X_train, X_test, y_train, y_test = train_test_split(
    df.drop('label', axis=1), df['label'],
    test_size=0.33, random_state=17)

X_train.sample(3)

	accountAgeDays	numItems	localTime	paymentMethodAgeDays	paymentMethod_creditcard	paymentMethod_paypal	paymentMethod_storecredit
27527	2000	1	5.017904	886.436111	1	0	0
5915	915	1	4.921349	345.954861	0	1	0
27704	470	1	3.066058	322.661111	1	0	0

X_test.sample(3)

	accountAgeDays	numItems	localTime	paymentMethodAgeDays	paymentMethod_creditcard	paymentMethod_paypal	paymentMethod_storecredit
16391	160	1	5.034622	0.000000	1	0	0
28741	1991	1	4.057414	1393.980556	1	0	0
5400	566	1	4.895263	0.000000	1	0	0

y_train.sample(3)

13004    0
19215    0
25897    0
Name: label, dtype: int64

y_test.sample(3)

5315     0
26059    1
22026    0
Name: label, dtype: int64

# Initialize and train classifier model
clf = LogisticRegression().fit(X_train, y_train)

# Make predictions on test set
y_pred = clf.predict(X_test)

y_pred

array([0, 0, 0, ..., 0, 0, 0])

# Compare test set predictions with ground truth labels
accuracy_score(y_pred, y_test)

1.0

print(confusion_matrix(y_test, y_pred))

[[12753     0]
 [    0   190]]