April 16, 2023

Introduction

This is a kaggle competition to predict which passengers are transported to an alternate dimension. The data source is from https://www.kaggle.com/competitions/spaceship-titanic

I blended predictive results using Vote Classifier from two algorithms – LGBM, and CatBoost – to maximize the accuracy of the prediction.

Achievement

• R-squared for test dataset after submission was 0.80897
• Rank: 133/2504 (Top 5% as of April 17, 2023)

Datasets

• 12 variables for each passenger and a column for successfully transported or not (y variable). Total 8,693 rows for training set, and 4,277 rows for test set.

• After data cleaning, I kept all rows and imputed missing values.

Language and libraries

Language: Python

Libraries :

• Data Manipulation: pandas, numpy
• Data Visualization: matplotlib, seaborn, msno
• Machine Learning: sklearn, xgboost, lightgbm, catboost

Data Preprocessing 1

• Insights :
  • Missing data is not in order