This is my first try on kaggle!
- The competition name is Santander Customer Transaction Prediction. It aims to identify which customers will make a specific transaction in the future, irrespective of the amount of money transacted.
- The competition time is from February 13 to April 10,2019
- Week 1
- Extract basic features
- Ready for initial model training.
- Week 2-4
- Learn the existing kernels and tools such as xgboost, lightgbm.
- Learn and implement my first model.
- Traning on the data set and submit my first outcome.
- Week 5-6
- Learn and apply feature engineering in my model.
- Week 7-8
- Learn and use other possible ways to improve my result.
The dataset is from kaggle. You can download from here. Also if you want to run my notebook file, make sure the working path is correct.
- My final ranking is 2331/8802, and my code and report is here.
- If you're looking for the best models. Here are the top solutions:1st place solution,2nd place solution.
I'm new to machine learning so the task might sound difficult to me. But the important thing is I want to improve my python coding skill and gain lots of information in machine learning through this competition. Although I've spent much time making little progress on my final result, I feel satisfied for what I've learnt from kaggle community.
See the LICENSE file for rights and limitations (MIT).