UCL Machine Learning dataset - BlogFeedback

Overview

This project performs a feature and model selection on the UCI machine learning BlogFeedback dataset. The methods used include ridge, lasso, and elastic net regressions. Given the performance metrics, 42 features are selected from 280 features via Lasso regression.

Data Description

Detailed documentation and data source are available here UCL - BlogFeedback Data Set

The dataset include 281 variables (280 features and 1 target variable). The data-attribute-description.md describe the names of the features which correspond to the columns of the dataset.

The original dataset includes one train set, blogData_train.csv and multiple small test set. I combined all the test set into one test set called test.csv. All analysis are based on the blogData_train.csv and test.csv.

Procedure of analysis

To reproduce the whole analysis, you need to clone the repo into your local machine.

Then open the jupyter notebook in src/analysis-script.ipynb, and run all from top to bottom.

Final report

The final report can be found in the doc folder report

Dependencies