Disco CLI

:fire: Generate recommendations from CSV files

• Supports user-based and item-based recommendations
• Works with explicit and implicit feedback
• Uses high-performance matrix factorization

Also available for Rust and Ruby

Installation

Download the latest version:

• Linux - x86_64
• Mac - x86_64 or arm64
• Windows - x86_64

You can also install it with Homebrew:

brew install ankane/brew/disco

or Cargo:

cargo install disco-cli

Quickstart

Download the MovieLens 100k dataset and generate item-based recommendations

disco download movielens-100k
disco item-recs movielens-100k.csv output.csv --factors 20
grep "^Star Wars" output.csv

How to Use

Data

Create a CSV file with your data. If users rate items directly, this is known as explicit feedback. The CSV should have three columns: user_id, item_id, and rating.

user_id,item_id,rating
1,post1,5
1,post2,3.5
2,post1,4

If users don’t rate items directly (for instance, they’re purchasing items or reading posts), this is known as implicit feedback. Use value instead of rating and a value like number of purchases, number of page views, or just 1.

user_id,item_id,value
1,post1,1
1,post2,1
2,post1,1

Each user_id/item_id combination should only appear once.

User-based Recommendations

Generate user-based recommendations - “users like you also liked”

disco user-recs data.csv output.csv

This creates a CSV with user_id, recommended_item_id, and score columns.

Item-based Recommendations

Generate item-based recommendations - “users who liked this item also liked”

disco item-recs data.csv output.csv

This creates a CSV with item_id, recommended_item_id, and score columns.

Similar Users

Generate similar users

disco similar-users data.csv output.csv

This creates a CSV with user_id, similar_user_id, and score columns.

Algorithms

Disco uses high-performance matrix factorization.

• For explicit feedback, it uses the stochastic gradient method with twin learners
• For implicit feedback, it uses the conjugate gradient method

Specify the number of factors and iterations

disco ... --factors 8 --iterations 20

Options

Specify the number of recommendations for each user or item

disco ... --count 10

Datasets

Download a dataset

disco download movielens-100k

Supported datasets are:

• movielens-100k
• movielens-1m
• movielens-25m
• movielens-latest-small
• movielens-latest

History

View the changelog

Contributing

Everyone is encouraged to help improve this project. Here are a few ways you can help:

• Report bugs
• Fix bugs and submit pull requests
• Write, clarify, or fix documentation
• Suggest or add new features

To get started with development:

git clone https://github.com/ankane/disco-cli.git
cd disco-cli
cargo run