# An Introduction to Redis-ML (Part Four)

Aug 30, 2017 · 3 min read

This post is part four of a series of posts introducing the Redis-ML module. The first article in the series can be found here.

In this post, we’re going to look at the matrix operations provided by the Redis-ML module and show some examples of how to process Matrix data using the Redis database.

Technical Requirements

The example code in this post is written in Python and requires a Redis instance running Redis-ML. The instructions for setting up Redis can be found in either part one or part two of this series.

Matrices in Redis

Matrix multiplication diagram from the Wikimedia Matrix Multiplication article

Matrices are common in machine learning, statistics, finance and a host of other domains, so they were a natural addition to Redis. The Redis-ML module adds matrices as a native Redis data type. It also provides mathematical operations that combine matrices to create new values.

Reading and writing matrix values is performed through the ML.MATRIX.SET and the ML.MATRIX.GET commands which have the following syntax:

`ML.MATRIX.SET key n m entry11 .. entrynmML.MATRIX.GET key`

When working with the Redis-ML module, remember that commands use row-major format. Multiplication and addition are supported by the ML.MATRIX.MULTIPLY and the ML.MATRIX.ADD commands.

`ML.MATRIX.MULTIPLY a b resultML.MATRIX.ADD a b result`

These commands combine two matrices that are already in Redis and store the result in a new key.

If I wanted to compute a basic Matrix equation such as y = Ax + b using the Redis-ML module, I would enter the following commands into the Redis CLI:

`127.0.0.1:6379> ML.MATRIX.SET a 3 3 4 0 0 0 2 0 0 0 1127.0.0.1:6379> ML.MATRIX.SET x 3 1 1 4 1127.0.0.1:6379> ML.MATRIX.SET b 3 1 1 1 1127.0.0.1:6379> ML.MATRIX.MULTIPLY a x tmp127.0.0.1:6379> ML.MATRIX.ADD tmp b y127.0.0.1:6379> ML.MATRIX.GET y1) (integer) 32) (integer) 13) "5"4) "9"5) "2"`

Redis returns the result to our client with the shape of the matrix (in this case 3 rows and 1 column) followed by each of the elements of the matrix in row-major order.

I could also compute the Matrix equation A’ = cA (where c is a scalar value) using the following code:

`127.0.0.1:6379> ML.MATRIX.SET a 3 3 4 0 0 0 2 0 0 0 1127.0.0.1:6379> ML.MATRIX.SCALE a 3127.0.0.1:6379> ML.MATRIX.GET a 1) (integer) 3 2) (integer) 3 3) "12" 4) "0" 5) "0" 6) "0" 7) "6" 8) "0" 9) "0"10) "0"11) "3"`

Matrices are used for a wide range of applications, from linear transforms to representing multivariate probability distributions. In the next post, we’ll look at decision trees and random forests, two additional classification models supported by Redis.

Please connect with me on twitter (@tague) if you have questions regarding this or previous posts in the series.

Explore Redis Enterprise at our brand new websiteredislabs.com — and experience the fastest and most prolific database in the world!

This blog was originally posted at: https://redislabs.com/blog/introduction-redis-ml-part-four/

If you would like to view past Redis Watch articles, or to subscribe to Redis Watch, please visit us here: https://redislabs.com/resources/redis-watch-archive/

Written by