Overview
Learn how to achieve a top 1% solution in the Santander Kaggle Transaction Competition without using ensemble methods. Follow along as the video guides you through obtaining the data, creating a simple neural network baseline, understanding and analyzing the dataset, modifying the neural network architecture, implementing feature engineering techniques, and making final improvements to achieve a high-scoring submission. Gain insights into effective problem-solving strategies for machine learning competitions and practical tips for improving model performance.
Syllabus
- Introduction to competition
- Get data
- Simple NN baseline
- First results 0.86 score
- Understanding the data
- Modifying our NN
- Improvement to baseline
- Feature engineering
- Modifying our NN v2
- Final result and submission
- Ending
Taught by
Aladdin Persson