marikalion/default-credit-card-prediction

Machine Learning Project for predicting Credit Card Defaults

Default-Credit-Card-Prediction

Machine Learning Project for predicting Credit Card Defaults.

Features:

• Dataset Loading
• Feature Assessment/Visualization:
  • Normalized Histogram Distribution
  • Box Plots
  • Pairwise Relationships
  • Empirical Cumulative/Standard Density Functions
  • Pearson Correlation
  • 2D PCA
  • 2D LDA
• Preprocessing:
  • Standardization
  • Scaling
  • Normalization
  • Dataset Balancing
• Feature Selection:
  • (Filter) Information Gain
  • (Filter) Gain Ratio
  • (Filter) Chi-squared Test
  • (Filter) Kruskal-Wallis Test
  • (Filter) Fisher Score
  • (Filter) Pearson Correlation (Feature-Feature, Feature-Class)
  • (Filter) mRMR
  • (Filter) Area Under the Curve (AUC)
  • (Wrapper) Sequential Forward/Backward Selection
  • (Wrapper) Recursive Feature Elimination
• Feature Reduction:
  • Principal Component Analysis (PCA)
  • Fisher's Linear Discriminant Analysis (LDA)
• Classification:
  • Minimum Distance Classifier
  • k-Nearest-Neighbors (kNN)
  • Naive Bayes
  • Support Vector Machines (SVM)
  • Decision Tree (CART)
  • Random Forest
• Evaluation:
  • Stratified K-folds Cross Validation
  • Receiver Operating Characteristic (ROC) Curves
  • Precision-Recall Curves

####Requirements:####

• python 2.x
• scikit-learn (http://scikit-learn.org/stable)
• sciPy (http://www.scipy.org)
• numPy (http://www.numpy.org)
• matplotlib (http://matplotlib.org)
• mlxtend (http://rasbt.github.io/mlxtend)
• UnbalancedDataset (https://github.com/shubhabrataroy/UnbalancedDataset)
• pandas (http://pandas.pydata.org)
• seaborn (https://stanford.edu/~mwaskom/software/seaborn)
• PyQT5 (https://pypi.python.org/pypi/PyQt5)

####Usage:####

usage: python default_predictor.py

####Examples:####