MY PAYPAL

4 min readDec 9, 2022

Hello everyone.Today’s topic is to build a fraud detection model using the provided data set and increase the revenue from transaction fees.

**Today I am in a different image. We will not pay attention.🤫**

I’m glad to see you all again. Here we go🚀

1>>Of course, we can start by simply checking the dataset

**Color is positive_green and negative_red**

2>>Statistical summary of all the numerical variables in dataset

3>>Company’s compaingn ids’ composition

**Evidently, data is highly imbalanced. Only 0.173% transactions are fradulent.**

4>>Visualizing distributions of all numerical variables

5>>Time

count    284807.000000
mean         26.337183
std          13.191152
min           0.000000
25%          15.055972
50%          23.525556
75%          38.700139
max          47.997778

Fraudulent transaction are more evenly distributed than non-fraudulent ones. It looks like fraudulent transactions were relatively lesser at higher time(the seconds elapsed between each transaction and the first transaction)

6>>Distribution of Transaction Amount

7>>Average transaction Amount for the two classes

8>>Distribution of transaction amount for the two classes

**It is evident that the average fraudulent transaction amount is greater than the average non-fraudulent transaction amount.**

9>>Correlation Matrix

There are no multi-colinearity in data. There are no notable correlations between variables. However, we observe some positive correlation between V7 and Amount, and V20 and Amount, and some negative correlation is observed between V2 and Amount, and V3 and Time.

Get Burxon_coder1’s stories in your inbox

Join Medium for free to get updates from this writer.

Remember me for faster sign in

**The variables appear to be correlated to some extent. Moving forward to Modelling where we can eliminate these features based on p-values.**

Data Modelling🔎

Let’s groupby our data by “class” first.

1_Model>>DecisionTreeClassifier

precision    recall  f1-score   support

           0       1.00      1.00      1.00     85295
           1       0.81      0.74      0.77       148

    accuracy                           1.00     85443
   macro avg       0.90      0.87      0.89     85443
weighted avg       1.00      1.00      1.00     85443    

                       << 0.9992509626300574 ✅ >>

2_Model<<SVC

              precision    recall  f1-score   support

           0       1.00      1.00      1.00     85295
           1       0.97      0.57      0.72       148

    accuracy                           1.00     85443
   macro avg       0.98      0.79      0.86     85443
weighted avg       1.00      1.00      1.00     85443  

                      << 0.9992275552122467 ✅ >>

3_Model>>XGBClassifier

              precision    recall  f1-score   support

           0       1.00      1.00      1.00     85295
           1       0.91      0.72      0.80       148

    accuracy                           1.00     85443
   macro avg       0.95      0.86      0.90     85443
weighted avg       1.00      1.00      1.00     85443 
                       
                      << 0.9993914071369217 ✅ >>

4_Model>>RandomForestClassifier

       precision    recall  f1-score   support

           0       1.00      1.00      1.00     85295
           1       0.95      0.72      0.82       148

    accuracy                           1.00     85443
   macro avg       0.98      0.86      0.91     85443
weighted avg       1.00      1.00      1.00     85443

                      << 0.9994499256814484 ✅ >>

5_Model>>LogisticRegression

         precision    recall  f1-score   support

           0       1.00      1.00      1.00     85295
           1       0.88      0.63      0.73       148

    accuracy                           1.00     85443
   macro avg       0.94      0.81      0.87     85443
weighted avg       1.00      1.00      1.00     85443

                      << 0.999204147794436 ✅ >>

Classification summary 🏁

1_DecisionTreeClassifier << 0.9992509626300574 ✅ >>
2_SVC                    << 0.9992275552122467 ✅ >>
3_XGBClassifier          << 0.9993914071369217 ✅ >>
4_RandomForestClassifier << 0.9994499256814484 ✅ >> 👈👈👈👈👈👈👈👍👍👍👍👍👍👍
5_LogisticRegression     << 0.9992041477944436 ✅ >>