kaggle-competitions

This is a block of models predictions and self implement libraries in PyTorch, in order to compete in the famous data scientist AI Kaggle competitions.

1 - House Price Prediction: LinearRegression model from sklearn used in this competition. KAGGLE, SCORE 0.55249 & RANKING 4862/5106*
2 - Titanic Survival Prediction: LinearRegression model from sklearn used in this competition. KAGGLE SCORE 0.0000 & RANKING 13632/13632*
3 - Stickers Sales Prediction: by selecting the best model with their respective MAPE loss, where we got models like LGBM, XGBOOST, CATBOOST from gradient boost and others that is used in this competition. KAGGLE SCORE 0.96936 & RANKING 1101/1297*
4 - Store Sales Prediction: by undestanding the dataset, we can infer that each dataset provided by the kaggle has its meaning with another dataset, so i did a little bit of feature engineering such as combining differents features or creating a new one. Then finally, we train with the RandomForest model that we got 2.0... of RMLE that is kind of good, but of course we can still improve it with more graphic understanding and model fine tuning. KAGGLE SCORE & RANKING 582/841*
5 - Spaceship Titanic Prediction: RF model with custom NAN value correction self implementation. KAGGLE SCORE 0.78793 & RANKING 1546/2312*.
6 - Coffee Bean Country Origin: Focusing in data cleaning and formatting, in order to get a good training for the model.
7 - Plant Species Classification: classify plants by taking a image from yolo model that detects pot of plants, and then predict the plant species with resnet18 pre-trained model.