There are notebooks from my participating in Kaggle competitions.
Churn prediction https://www.kaggle.com/competitions/advanced-dls-spring-2021
- finding missing data
- visualizing data
- applying linear models
- used CatBoost for my final submission, hyperparameter search was done
Classification images https://www.kaggle.com/competitions/journey-springfield
- build simple CNN for baseline submission
- took pretrained VGG16 model. Freezed features extractor and trained classifier with different learning rates. Another experiment was with different learning rates for features extractor and classifier.
- took pretrained EfficientNet and fine-tuned for my task