Demonstration of the effect of synthetic features on linear models
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
data
README.md Updated readme for data source. Sep 1, 2013
harversine_distance_dataset_with_synthetic_features.png
harversine_distance_dataset_without_synthetic_features.png
stacked_generalization.py
try_linear_with_bruteforce_features.py
try_predicting_with_simple_synthetic_features.py Added one more example. Sep 1, 2013

README.md

Purpose: This script investigates whether or not synthetically constructed features will help make the linear model (LogisticRegression) predict better.

Author: Eric Chio "log0" im.ckieric@gmail.com

Summary: Our test aims to find out if artificial features could make the prediction more accurate. In our case, we found that an additional of 2 features or more (despite discovered via bruteforce manner) does increase the prediction capabilities.

Dataset is available here : http://archive.ics.uci.edu/ml/datasets/Vertebral+Column