Credit Card Fraud Detection¶
Jaime Avendaño
This is a classification notebook that will use a Naive Bayes classifier. The data has been created through a previous PCA transformation.
The data is also highly unbalanced.
In [3]:
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from matplotlib.colors import ListedColormap
import seaborn as sns
%matplotlib inline
In [57]:
from sklearn.preprocessing import LabelEncoder, MinMaxScaler
from sklearn.model_selection import train_test_split, GridSearchCV
from sklearn.naive_bayes import MultinomialNB, GaussianNB
from sklearn.metrics import classification_report, confusion_matrix
from imblearn.over_sampling import SMOTE
In [18]:
data = pd.read_csv('Course_Data\creditcard.csv')
#data = data.clean_names(case_type='snake')
data.head()
Out[18]:
| Time | V1 | V2 | V3 | V4 | V5 | V6 | V7 | V8 | V9 | ... | V21 | V22 | V23 | V24 | V25 | V26 | V27 | V28 | Amount | Class | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 0.0 | -1.359807 | -0.072781 | 2.536347 | 1.378155 | -0.338321 | 0.462388 | 0.239599 | 0.098698 | 0.363787 | ... | -0.018307 | 0.277838 | -0.110474 | 0.066928 | 0.128539 | -0.189115 | 0.133558 | -0.021053 | 149.62 | 0 |
| 1 | 0.0 | 1.191857 | 0.266151 | 0.166480 | 0.448154 | 0.060018 | -0.082361 | -0.078803 | 0.085102 | -0.255425 | ... | -0.225775 | -0.638672 | 0.101288 | -0.339846 | 0.167170 | 0.125895 | -0.008983 | 0.014724 | 2.69 | 0 |
| 2 | 1.0 | -1.358354 | -1.340163 | 1.773209 | 0.379780 | -0.503198 | 1.800499 | 0.791461 | 0.247676 | -1.514654 | ... | 0.247998 | 0.771679 | 0.909412 | -0.689281 | -0.327642 | -0.139097 | -0.055353 | -0.059752 | 378.66 | 0 |
| 3 | 1.0 | -0.966272 | -0.185226 | 1.792993 | -0.863291 | -0.010309 | 1.247203 | 0.237609 | 0.377436 | -1.387024 | ... | -0.108300 | 0.005274 | -0.190321 | -1.175575 | 0.647376 | -0.221929 | 0.062723 | 0.061458 | 123.50 | 0 |
| 4 | 2.0 | -1.158233 | 0.877737 | 1.548718 | 0.403034 | -0.407193 | 0.095921 | 0.592941 | -0.270533 | 0.817739 | ... | -0.009431 | 0.798278 | -0.137458 | 0.141267 | -0.206010 | 0.502292 | 0.219422 | 0.215153 | 69.99 | 0 |
5 rows × 31 columns
In [19]:
not_fraud = data[data.Class == 0]
fraud = data[data.Class == 1]
data.shape, not_fraud.shape, fraud.shape
Out[19]:
((284807, 31), (284315, 31), (492, 31))
EDA¶
In [11]:
data['class'].value_counts()
Out[11]:
0 284315 1 492 Name: class, dtype: int64
In [13]:
data.describe()
Out[13]:
| time | v1 | v2 | v3 | v4 | v5 | v6 | v7 | v8 | v9 | ... | v21 | v22 | v23 | v24 | v25 | v26 | v27 | v28 | amount | class | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| count | 284807.000000 | 2.848070e+05 | 2.848070e+05 | 2.848070e+05 | 2.848070e+05 | 2.848070e+05 | 2.848070e+05 | 2.848070e+05 | 2.848070e+05 | 2.848070e+05 | ... | 2.848070e+05 | 2.848070e+05 | 2.848070e+05 | 2.848070e+05 | 2.848070e+05 | 2.848070e+05 | 2.848070e+05 | 2.848070e+05 | 284807.000000 | 284807.000000 |
| mean | 94813.859575 | 3.919560e-15 | 5.688174e-16 | -8.769071e-15 | 2.782312e-15 | -1.552563e-15 | 2.010663e-15 | -1.694249e-15 | -1.927028e-16 | -3.137024e-15 | ... | 1.537294e-16 | 7.959909e-16 | 5.367590e-16 | 4.458112e-15 | 1.453003e-15 | 1.699104e-15 | -3.660161e-16 | -1.206049e-16 | 88.349619 | 0.001727 |
| std | 47488.145955 | 1.958696e+00 | 1.651309e+00 | 1.516255e+00 | 1.415869e+00 | 1.380247e+00 | 1.332271e+00 | 1.237094e+00 | 1.194353e+00 | 1.098632e+00 | ... | 7.345240e-01 | 7.257016e-01 | 6.244603e-01 | 6.056471e-01 | 5.212781e-01 | 4.822270e-01 | 4.036325e-01 | 3.300833e-01 | 250.120109 | 0.041527 |
| min | 0.000000 | -5.640751e+01 | -7.271573e+01 | -4.832559e+01 | -5.683171e+00 | -1.137433e+02 | -2.616051e+01 | -4.355724e+01 | -7.321672e+01 | -1.343407e+01 | ... | -3.483038e+01 | -1.093314e+01 | -4.480774e+01 | -2.836627e+00 | -1.029540e+01 | -2.604551e+00 | -2.256568e+01 | -1.543008e+01 | 0.000000 | 0.000000 |
| 25% | 54201.500000 | -9.203734e-01 | -5.985499e-01 | -8.903648e-01 | -8.486401e-01 | -6.915971e-01 | -7.682956e-01 | -5.540759e-01 | -2.086297e-01 | -6.430976e-01 | ... | -2.283949e-01 | -5.423504e-01 | -1.618463e-01 | -3.545861e-01 | -3.171451e-01 | -3.269839e-01 | -7.083953e-02 | -5.295979e-02 | 5.600000 | 0.000000 |
| 50% | 84692.000000 | 1.810880e-02 | 6.548556e-02 | 1.798463e-01 | -1.984653e-02 | -5.433583e-02 | -2.741871e-01 | 4.010308e-02 | 2.235804e-02 | -5.142873e-02 | ... | -2.945017e-02 | 6.781943e-03 | -1.119293e-02 | 4.097606e-02 | 1.659350e-02 | -5.213911e-02 | 1.342146e-03 | 1.124383e-02 | 22.000000 | 0.000000 |
| 75% | 139320.500000 | 1.315642e+00 | 8.037239e-01 | 1.027196e+00 | 7.433413e-01 | 6.119264e-01 | 3.985649e-01 | 5.704361e-01 | 3.273459e-01 | 5.971390e-01 | ... | 1.863772e-01 | 5.285536e-01 | 1.476421e-01 | 4.395266e-01 | 3.507156e-01 | 2.409522e-01 | 9.104512e-02 | 7.827995e-02 | 77.165000 | 0.000000 |
| max | 172792.000000 | 2.454930e+00 | 2.205773e+01 | 9.382558e+00 | 1.687534e+01 | 3.480167e+01 | 7.330163e+01 | 1.205895e+02 | 2.000721e+01 | 1.559499e+01 | ... | 2.720284e+01 | 1.050309e+01 | 2.252841e+01 | 4.584549e+00 | 7.519589e+00 | 3.517346e+00 | 3.161220e+01 | 3.384781e+01 | 25691.160000 | 1.000000 |
8 rows × 31 columns
In [24]:
data_cols = data.drop(columns=['Time', 'Amount', 'Class']).columns
data_cols
Out[24]:
Index(['V1', 'V2', 'V3', 'V4', 'V5', 'V6', 'V7', 'V8', 'V9', 'V10', 'V11',
'V12', 'V13', 'V14', 'V15', 'V16', 'V17', 'V18', 'V19', 'V20', 'V21',
'V22', 'V23', 'V24', 'V25', 'V26', 'V27', 'V28'],
dtype='object')
In [27]:
i = 0
fig, ax = plt.subplots(10,3,figsize=(20,40))
for col in data_cols:
i += 1
plt.subplot(10,3, i)
sns.kdeplot(data=fraud, x=col, bw_method=0.4, label="Fraud", shade=True, color="r", linestyle="--")
sns.kdeplot(data=not_fraud, x=col, bw_method=0.4, label="Non Fraud", shade=True, color= "y", linestyle=":")
plt.title(col, fontsize=12)
plt.show();
Data Cleaning¶
- Drop Time
- MinMaxScaler on all features
In [36]:
X = data.drop(columns=['Time', 'Class'])
y = data.Class
In [37]:
scaler = MinMaxScaler()
num_features = X.columns
scaler.fit(X)
X = scaler.transform(X)
X[:2]
Out[37]:
array([[9.35192337e-01, 7.66490419e-01, 8.81364903e-01, 3.13022659e-01,
7.63438735e-01, 2.67668642e-01, 2.66815176e-01, 7.86444198e-01,
4.75311734e-01, 5.10600482e-01, 2.52484319e-01, 6.80907625e-01,
3.71590602e-01, 6.35590530e-01, 4.46083696e-01, 4.34392391e-01,
7.37172553e-01, 6.55065861e-01, 5.94863228e-01, 5.82942230e-01,
5.61184389e-01, 5.22992116e-01, 6.63792975e-01, 3.91252676e-01,
5.85121795e-01, 3.94556792e-01, 4.18976135e-01, 3.12696634e-01,
5.82379309e-03],
[9.78541955e-01, 7.70066651e-01, 8.40298490e-01, 2.71796491e-01,
7.66120336e-01, 2.62191698e-01, 2.64875439e-01, 7.86298353e-01,
4.53980968e-01, 5.05267346e-01, 3.81187722e-01, 7.44341569e-01,
4.86190176e-01, 6.41219007e-01, 3.83839664e-01, 4.64105178e-01,
7.27793983e-01, 6.40681094e-01, 5.51930422e-01, 5.79529753e-01,
5.57839915e-01, 4.80236960e-01, 6.66937823e-01, 3.36439996e-01,
5.87290252e-01, 4.46012969e-01, 4.16345145e-01, 3.13422663e-01,
1.04705276e-04]])
In [38]:
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
Model Training and Testing¶
In [39]:
NB_classifier = MultinomialNB()
NB_classifier.fit(X_train, y_train)
Out[39]:
MultinomialNB()
In [40]:
y_pred = NB_classifier.predict(X_test)
y_pred
Out[40]:
array([0, 0, 0, ..., 0, 0, 0], dtype=int64)
In [41]:
cm = confusion_matrix(y_test, y_pred)
sns.heatmap(cm, annot=True, fmt='d')
Out[41]:
<AxesSubplot:>
In [48]:
print(classification_report(y_test, y_pred))
precision recall f1-score support
0 1.00 1.00 1.00 56864
1 0.00 0.00 0.00 98
accuracy 1.00 56962
macro avg 0.50 0.50 0.50 56962
weighted avg 1.00 1.00 1.00 56962
C:\Users\jaime.avendano\Anaconda3\lib\site-packages\sklearn\metrics\_classification.py:1221: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior. _warn_prf(average, modifier, msg_start, len(result))
In [44]:
Gaus_classifier = GaussianNB()
Gaus_classifier.fit(X_train, y_train)
Out[44]:
GaussianNB()
In [46]:
y_pred_gaus = Gaus_classifier.predict(X_test)
y_pred_gaus
Out[46]:
array([1, 0, 0, ..., 0, 0, 0], dtype=int64)
In [47]:
cm = confusion_matrix(y_test, y_pred_gaus)
sns.heatmap(cm, annot=True, fmt='d')
Out[47]:
<AxesSubplot:>
In [49]:
print(classification_report(y_test, y_pred_gaus))
precision recall f1-score support
0 1.00 0.98 0.99 56864
1 0.06 0.82 0.11 98
accuracy 0.98 56962
macro avg 0.53 0.90 0.55 56962
weighted avg 1.00 0.98 0.99 56962
Handling unbalanced data¶
In [58]:
X_resample, y_resample = SMOTE().fit_sample(X,y.values.ravel())
X_resample.shape, y_resample.shape
Out[58]:
((568630, 29), (568630,))
In [62]:
y.value_counts(), np.bincount(y_resample)
Out[62]:
(0 284315 1 492 Name: Class, dtype: int64, array([284315, 284315], dtype=int64))
In [59]:
X_train, X_test, y_train, y_test = train_test_split(X_resample, y_resample,
test_size=0.2, random_state=42)
In [63]:
NB_classifier = MultinomialNB()
NB_classifier.fit(X_train, y_train)
Out[63]:
MultinomialNB()
In [64]:
y_pred = NB_classifier.predict(X_test)
y_pred
Out[64]:
array([1, 1, 0, ..., 1, 1, 0], dtype=int64)
In [65]:
cm = confusion_matrix(y_test, y_pred)
sns.heatmap(cm, annot=True, fmt='d')
Out[65]:
<AxesSubplot:>
In [66]:
print(classification_report(y_test, y_pred))
precision recall f1-score support
0 0.81 1.00 0.90 56750
1 1.00 0.77 0.87 56976
accuracy 0.88 113726
macro avg 0.91 0.89 0.88 113726
weighted avg 0.91 0.88 0.88 113726
In [67]:
Gaus_classifier = GaussianNB()
Gaus_classifier.fit(X_train, y_train)
Out[67]:
GaussianNB()
In [68]:
y_pred_gaus = Gaus_classifier.predict(X_test)
y_pred_gaus
Out[68]:
array([1, 1, 0, ..., 1, 1, 0], dtype=int64)
In [69]:
cm = confusion_matrix(y_test, y_pred_gaus)
sns.heatmap(cm, annot=True, fmt='d')
Out[69]:
<AxesSubplot:>
In [70]:
print(classification_report(y_test, y_pred_gaus))
precision recall f1-score support
0 0.86 0.98 0.92 56750
1 0.97 0.85 0.91 56976
accuracy 0.91 113726
macro avg 0.92 0.91 0.91 113726
weighted avg 0.92 0.91 0.91 113726
In [71]:
y_predict = Gaus_classifier.predict(X)
cm = confusion_matrix(y, y_predict)
sns.heatmap(cm, annot=True, fmt='d')
Out[71]:
<AxesSubplot:>
In [72]:
print(classification_report(y, y_predict))
precision recall f1-score support
0 1.00 0.98 0.99 284315
1 0.06 0.86 0.11 492
accuracy 0.98 284807
macro avg 0.53 0.92 0.55 284807
weighted avg 1.00 0.98 0.99 284807