2017-12-15

Kaggle workshop

Link to the material on github.

Overview

I created this material for a pre-christmas event at our department. The goal was to enable even complete ML beginners (all had an engineering major though) to participant in a kaggle competition. The workshop was divided into two parts. In the first part I went through some Machine Learning basics using the titanic dataset from kaggle. This took about 3-4 hours. Here is the outline:

  1. Machine Learning introduction
  2. Linear models
  3. Pandas introduction
  4. Exploratory Data Analysis (EDA) using Seaborn
  5. Feature engineering
  6. Model evaluation and cross validation
  7. Regularization
  8. Decision Trees
  9. K-nearest Neighbor
  10. Hyperparameter optimization
  11. Ensemble methods (bagging, boosting)

In the second part we built teams and worked on the [housing price] (https://www.kaggle.com/c/house-prices-advanced-regression-techniques) challenge on kaggle. I created three different starter kits (all included in the repo) so that even people with no ML background could get to a decent solution.

Feel free to use it in whatever way you want!

comments powered by Disqus