MyMediaLite: How to use a recommender in F#

News

MyMediaLite 3.11 has been released.

On this page, we show how to first set up recommenders in F#, and then use them to make predictions.

To use the examples on this page, download the MovieLens 100k ratings dataset from the GroupLens Research website and unzip it.
Of course you can also use your own data ;-)

Rating Prediction

open System
open MyMediaLite.IO
open MyMediaLite.RatingPrediction
open MyMediaLite.Eval

(* load the data *)
let train_data = RatingData.Read "u1.base"
let test_data  = RatingData.Read "u1.test"

(* set up the recommender *)
let recommender = new UserItemBaseline(Ratings=train_data)
recommender.Train()

(* measure the accuracy on the test data set *)
let result = recommender.Evaluate(test_data)
Console.WriteLine(result)

(* make a prediction for a certain user and item *)
let prediction = recommender.Predict(1, 1)
Console.WriteLine(prediction)

Item Prediction from Positive-Only Feedback

open System
open MyMediaLite.IO
open MyMediaLite.ItemRecommendation
open MyMediaLite.Eval

(* load the data *)
let train_data = ItemData.Read "u1.base"
let test_data = ItemData.Read "u1.test"

(* set up the recommender *)
let recommender = new UserKNN(K=20u, Feedback=train_data)
recommender.Train()

(* measure the accuracy on the test data set *)
let result = recommender.Evaluate(test_data, train_data)
Console.WriteLine(result)

(* make a prediction for a certain user and item *)
let prediction = recommender.Predict(1, 1)
Console.WriteLine(prediction)