A LITTLE TASTE. A BETTER WATCHLIST.
Your next
movie night.
Start with movies you know.
Find something you’ll love.
What have you seen?
Add a few favorites and a few misses. Three or more is a great start.
Try Tanner’s 28 available favorites, or build your own list below.
Type a title, then select a movie from the dropdown.
Like 0
The ones you’d watch again.
Dislike 0
The ones that weren’t for you.
Your picks stay in this browser. No account needed.
Made for your taste.
Your recommendations will appear here.
Good movies start with you.
Add a movie to either list and we’ll find your next watch. Keep adding picks to fine-tune the results.
Behind the recommendations
A 48-factor truncated SVD model learns patterns from the user–movie rating matrix in MovieLens 32M. Your likes (5 stars) and dislikes (1 star) fit a personal profile using regularized least squares. Each recommendation uses up to five closest liked titles and five closest disliked titles. Ranking sums their similarity weights and subtracts the dislike weights; it does not average them or sort by contributor count. Similarities at or below 0.25 contribute nothing. Expand a result to see the contributing likes and their weights. Your existing picks are excluded.
The entire catalog is searchable. Recommendations require at least 100 ratings for a more reliable signal. Movies without ratings cannot be used as preferences. The dataset covers ratings through October 2023, so newer releases are not included. Computation runs on your device. Tanner’s favorites button loads both tiers as Likes; Next Goal Wins (2023) and Dune: Part Two are unavailable.
Dataset: F. Maxwell Harper and Joseph A. Konstan. 2015. The MovieLens Datasets: History and Context. This independent, noncommercial experiment is not endorsed by GroupLens or the University of Minnesota. Data and derived model are subject to the MovieLens license. Training and validation details · Source code