Credit-Card-Fraud-Detection

In this project we have use various predictive models to see how accurate they are in detecting whether a transaction is a normal payment or a fraud.

Our Goals:

• Understand the little distribution of the "little" data that was provided to us.
• Create a 50/50 sub-dataframe ratio of "Fraud" and "Non-Fraud" transactions. (NearMiss Algorithm)
• Determine the Classifiers we are going to use and decide which one has a higher accuracy.
• Create a Neural Network and compare the accuracy to our best classifier.
• Understand common mistaked made with imbalanced datasets.

Challenges

• the most important challenge here is to handle data imbalance. the number of Fraud transection is much less than comparing to Non-fraud transection. In ML, data balancing is crucial because it has a huge impact in model training. In this project, we have implemeted several data balancing method to handle this cahllenges and improve model perfromance.

Things we have done here:

• Done Exploratory Data Abalysis
  • find relation between data
  • find their distribution and other related statistics
  • etc.
• Done Feature Selection:
  • t-sne + pca
  • Anova
• Handele Data Imbalance:
  • Random Undesampling
  • SMOTE
  • Combination of SMOTE + UnderSampling
• Apply Machine Learning Model:
  • KNN
  • Logistic Regreesion
  • Decision Tree
  • Random Forest
  • Nweural Network
• Evalue Model using
  • matirces such as accuracy,precision etc.
  • Confusion matrices