DataScienceTutorial

Bunch of Notebooks filled with information about DS libraries, methods and implementations studied from DeepLearningSchool and other sources

1. Algoritms:

1.1 Linear Regression:

Basic implementation in LinearRegression.ipynb. Also i implemented Ridge and Lasso regularization, random Batch without division on epochs

1.2. Logistic Regression:

Basic implementation in LogisticRegression.ipynb

1.3. Linear Models with Bagging:

Basic implementation in bagging.py

1.4. Gradient Boosting:

Basic implementation in boosting.py

2. Real datasets approach:

1. Titanic

Not the best, but at least existing solution to Titanic Kaggle problem. Solution based on Logistic Regression with ElasticNet. Coefficients were find through self-implemented values_grid, so the result may be improved with change of model, on that ML based, or with more detail work with data. Maybe 'Rich' feature isn't that good, as it was predicted. Or maybe, i shouldn't thrown away all NaN data ;)))

Implementation in kaggle\titanic\main.ipynb, solution in submit.csv

2. Prediction of the Churn of clients

Algorithm based on comparison of Random Forest, CatBoost and LogisticRegression

Performance may be increased with rebalancing the train.csv

Implementation in kaggle\PredictionOfTheChurnOfClients\main.ipynb, solution in submit.csv