Machine Learning: Make Your Own Recommender System

5

 

 

 

DATA REDUCTION

 

Recommender systems rely on data as input, and this can be collected through an assortment of channels including sign-up forms, web crawling, logs of online user behavior, IP and geographic tracking, and many other channels.

There are two major categories of input data: structured data and unstructured data. Structured data is information that resides in a fixed field within a record or file. This generally comprises information stored in rows and columns with a defined schema (a blueprint of how the database is constructed). Examples of structured data include event registration information stored in the rows and columns of a spreadsheet or user profiles stored in a relational database, including users’ personal ...