Feature hashing

Feature hashing

Introduction

Published in Max Kuhn, Kjell Johnson, Feature Engineering and Selection, 2019

Chapter 5 will focus on approaches for encoding discrete, or categorical, predictors. Here we will summarize standard techniques for representing categorical predictors. Feature engineering methods for categorical predictors such as feature hashing are introduced as a method for using existing information to create new predictors that better uncover meaningful relationships. This chapter will also provide guidance on practical issues such as how to handle rare levels within a categorical predictor and the impact of creating dummy variables for tree and rule-based models. Date-based predictors are present in many data sets and can be viewed as categorical predictors.

The Bitwise Hashing Trick for Personalized Search

View Article

Journal Information

Published in Applied Artificial Intelligence, 2019

Braddock Gaskill

A common method for representing and comparing items such as documents or listing titles is with a vector created by feature hashing aka “the Hashing Trick” (Attenberg, Weinberger, and Dasgupta et al. 2009; Weinberger et al. 2009). The vector is initialized to zero. Each feature is hashed to an index number modulo the vector length. The vector element at the index position is then incremented, or in some implementations decremented based on a second sign hash. Items are then compared pairwise using a similarity function such as the cosine between the vectors.

Explore chapters and articles related to this topic

Introduction

The Bitwise Hashing Trick for Personalized Search