Project - Credit Score classification

This Project is created as part of introduction to machine learning course included in Data Science Studies. We are going to use the dataset from Kaggle: Credit Score. Project is going to be done by a team of 3 people. Throughout the project we are going to be validated by another team.

Project team:

• Norbert Frydrysiak
• Anna Ostrowska
• Dominika Gimzicka

Validators team:

• Natalia Choszczyk
• Karolina Dunal

First part of the project - Exploratory Data Analysis

• The dataset is going to be divided into 4 parts: credit_score_train, credit_score_test, credit_score_valid, credit_score_validators.
• One part is only available for the Validators team.
• Each team member takes dataset and performs EDA on it separately.
• The results are then compared and discussed in the team.
• The final version of the EDA is going to be put in "EDA/final" folder.
• Validators are going to check the final version of the EDA.
• Project team is going to think of feedback from the validators and improve the final version of the EDA.
• Whole process can be repeated or changed if needed.

Second part of the project - Feature engineering and First models

• After EDA we know that there are no missing values in the dataset.
• We are going to ordinally encode "CAT_GAMBLING" column.
• We try to deal with outliers in two ways:
  • We are going to manually remove some of the outliers.
  • We are going to compare it with PYOD library functions.
• We are going to transform continuous variables using Box-Cox transformation and StandardScaler.
• We might try to use for instance PCA to reduce the number of features.
• We are going to compare the results of the models using different methods of dealing with outliers and different methods of feature engineering.
• We are going to use MANY different models to compare them.
• We will look at hiperparameters and try to optimize them.
• We are going to use cross-validation to compare the models.
• We are going to use different metrics to compare the models.
• Validators will check the results in feature_engineering/final.ipynb file.
• Project team is going to think of feedback from the validators and improve the final version of the feature engineering and first models.

Third part of the project - Final models and conclusions

• Files from this part can be found in "final_models" folder.
• We have done lots of things, I might describe them later.
• In folder "library" we have our own library with functions that we have used in the project.

Final reports

• Reports are written in Polish.
• They can be found in "reports" folder.
• We have report from our whole project and from each part of the project.
• We have also report about our validator work for the other team.

Note for every contributor:

• All files names should contain underscore instead of spaces. It is a good practice, because it is better for UNIX or UNIX-like systems.
• You can use Polish language, because this is pretty language, but I will try to use English as much as possible.