Wadood Alam

18th March 2024

AI 539 Final Project: Credit Score Evaluation

https://github.com/wadoodalam/Credit-Score-Prediction/blob/main/10kData.csv

• Python
• NumPy
• Pandas
• train_test_split
• StratifiedGroupKFold
• Scikit-learn
• accuracy_score
• confusion_matrix
• HistGradientBoostingClassifier
• DummyClassifier
• scipy.stats.mstats: winsorize
• time
• imblearn.under_sampling: RandomUnderSampler
• compute_sample_weight
• re(regex)
• matplotlib.pyplot

1. Install the required dependencies using pip
2. Ensure Dataset(train.csv) is contained in the same directory
3. Run the program using the command data_profile.py
4. The program will print Total number of missing values, Total number of missing values, Number of numerical features:
5. The program will generate 4 csv files, 1 xlsx file, and 5 png files
6. 10kdata.csv: The cleaned version of dataset containing 10,000 rows and 34 features
7. Cat_profile.csv: The data profile for the 4 relevant categorical features
8. correlation_matrix.xlsx: Correlation matrix
9. profile.csv: The data profile for numeric features
10. Credit_Mix.png: Representation of Credit_Mix feature
11. Credit_Score.png: Representation of Credit_Score feature
12. missing_values.png: Representation of the name of the feature with missing value and the number of values missing for each feature
13. Month.png: Representation of Month feature
14. Payment_Behaviour.png: Representation of Payment_Behaviour feature

1. Install the required dependencies using pip(if not installed previously)
2. Ensure Dataset(10kdata.csv) is contained in the same directory
3. Run the program using the command train_eval.csv
4. The program will output 3 dictionaries for accuracies, 3 dictionaries for runtime, 3 dictionaries for confusion matrices
5. The directories for confusion matrices and accuracies follow the following format: {'Strategy name':[Train-test-split, Stratified Group-wise Cross-Validation],...}
6. The runtime dictionary will follow the following format: Run Time: {'Strategy name': [],...}
7. The program will generate 1 csv file called outliers.csv to visualize outliers

1. train.csv
2. 10kdata.csv
3. Cat_profile.csv
4. correlation_matrix.xlsx
5. profile.csv
6. Credit_Mix.png
7. Credit_Score.png
8. missing_values.png
9. Month.png
10. Payment_Behaviour.png