Tabular Playground Series(TPS) July 2021

Published in Kaggle, 2021

Competition: https://www.kaggle.com/c/tabular-playground-series-jul-2021

Models used

Main Features

  • Feature creation
  • Feature selection
  • Log scaling outputs
  • K-folds training
  • AutoML for optimization - tried PyCaret and AutoGluon

  • Smart ensembling

File structure

├── input                
│   ├── train.csv
│   ├── test.csv
│   ├── submission_sample.csv
├── model_01
│   ├── 
├── model_02
│   ├── 
├── model_03
│   ├── 
├── Images
│   ├── Leaderboard.png
│   ├── Pipeline.png
│   ├── My_Profile.png
├── Readme.md
├── EDA_analysis.ipynb
├── Feature_selection and PCA.ipynb
├── Train_&_Predict.ipynb
└── requirements.txt

Data Preparation

Pipeline launch example

Training: Inference: Submit:

Best experiments:

Picture of leaderboard

Final Submission

Private Leaderboard: