Sparse Models

Overview


Sparse models are models with a limited (small) number of factors. That is, each record in a dataset may have a large number of features, {% X_1, X_2,...,X_m %}. A model built from these features is a function
{% y = f(X_1, X_2,...,X_m) %}
A sparse model is one build from a reduced set of features, say {% X_{i_1}, X_{i_2},...,X_{i_k} %} where the {% i_j \in \{ 1,2,...,m \} %} and {% k \lt m %}

Topics


  • OLS Regression and L1 Regularization