Credit_card fraud transaction identification using Python
-
Data collection : [Link] (https://www.kaggle.com/datasets/kartik2112/fraud-detection?select=fraudTest.csv ) Dataset collected from kaggle.
-
Data handle : Pandas to read the dataset and check null values,categorical values fill with mode and numerical values fill with median.
-
Data split: Split the dataset to Features (input) and Target (output) for training and testing
-
Encoding: The category, merchant, State to be label encoding ,it was convert into numerical labels.
-
Scalling: Support vector, Naive bayes needed scalling then features standard scaler.
-
Algorithms: Using multiple algorithms for comparison
Logistic Regreesion, Kneighbours, Support vector, Naive Bayes, Decision Tree classifier, Random Forest classifier -
Save model: Models are saved in the pickle file for prediction.
-
Techniques: Python Jupyter Notebook
-
Data analysis : Performed EDA using Matplotlib and seaborn for visual insights.
-
Result: Random forest achived
accuracy 95%it perform well on fraud identification.