{"id":24607616,"url":"https://github.com/adrianmarino/recommendation-system-approaches","last_synced_at":"2026-05-04T00:31:15.349Z","repository":{"id":89013142,"uuid":"293566568","full_name":"adrianmarino/recommendation-system-approaches","owner":"adrianmarino","description":"Recommendation system approaches","archived":false,"fork":false,"pushed_at":"2022-03-01T15:10:02.000Z","size":35722,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":2,"default_branch":"master","last_synced_at":"2025-09-13T19:43:15.045Z","etag":null,"topics":["deep-learning","keras","modin","movielens","ray","recommender-system","spark","tensorflow"],"latest_commit_sha":null,"homepage":"","language":"Jupyter Notebook","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/adrianmarino.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2020-09-07T15:34:19.000Z","updated_at":"2022-02-27T15:44:06.000Z","dependencies_parsed_at":"2023-06-13T13:15:37.357Z","dependency_job_id":null,"html_url":"https://github.com/adrianmarino/recommendation-system-approaches","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/adrianmarino/recommendation-system-approaches","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/adrianmarino%2Frecommendation-system-approaches","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/adrianmarino%2Frecommendation-system-approaches/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/adrianmarino%2Frecommendation-system-approaches/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/adrianmarino%2Frecommendation-system-approaches/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/adrianmarino","download_url":"https://codeload.github.com/adrianmarino/recommendation-system-approaches/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/adrianmarino%2Frecommendation-system-approaches/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":32590100,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-03T22:12:39.696Z","status":"ssl_error","status_checked_at":"2026-05-03T22:09:10.534Z","response_time":103,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["deep-learning","keras","modin","movielens","ray","recommender-system","spark","tensorflow"],"created_at":"2025-01-24T17:39:07.903Z","updated_at":"2026-05-04T00:31:15.309Z","avatar_url":"https://github.com/adrianmarino.png","language":"Jupyter Notebook","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Recommendation system approaches\n\nImplementation, test and comparatives of different recommendation models (Collaborative filtering like).  \n \n# Models\n\n* **Embedding dot model**\n  * Learn both users and movie embedding.\n  * Dot product between these to predict user ratings.\n\n* **Embedding biases dot model**\n  * Add a bias to each user and movie. Similar to the bias in a fully-connected layer or the intercept in a linear model. It just provides an extra degree of freedom.\n  * Also pass dot product output through a sigmoid layer and then scaling the result using the min and max ratings in the data. This technique introduces a non-linearity into the output and results in a loss improvement.\n\n* **Embedding dense model**\n  * Instead of performing a dot product between users and movies, embedding adds a fully connected layer(dense) after input embedding layers.\n  * This technique is more flexible than the previous ones because it allows adding more feature columns as model input.\n\n* **User/Movie/Gender embedding dense model**\n  * Same **Embedding dense model** approach.\n  * Add genders features columns.\n  * Use a sigmoid layer and then scale the result using the min and max ratings in the data.\n\n* **Wide and deep model**\n  * A mix between two models, linear regression and **Embedding dense model**.\n  * This model learns to combine memorization and generalization like humans do.\n  * Liner model learning to memorize.\n  * Deep model learning to generalize.\n  * See: [Wide \u0026 Deep Learning: Better Together with TensorFlow](https://ai.googleblog.com/2016/06/wide-deep-learning-better-together-with.html)\n\n* **Deep Factorization Machine Model**\n  * **Pending**. See: [Deep Factorization Machine Model for CRT prediction](https://github.com/adrianmarino/deep-fm).\n\n# Notebooks\n\n* [Recommendation system: Approaches](https://github.com/adrianmarino/recommendation-system-approaches/blob/master/recommendation-system-comparatives.ipynb)\n* [Recommendation systems: Deep Model only](https://github.com/adrianmarino/recommendation-system-approaches/blob/master/deep-model-user-movie.ipynb)\n* User/Movie/Genders Deep Model\n  * [1. Dataset preprocecing](https://github.com/adrianmarino/recommendation-system-approaches/blob/master/user-movie-genres-model/1.input-data-building.ipynb)\n  * [2. Dataset preprocecing and train/validation/test split](https://github.com/adrianmarino/recommendation-system-approaches/blob/master/user-movie-genres-model/2.train-test-sets-building.ipynb)  \n  * [3.1. Model training/validation (Pandas Ray version)](https://github.com/adrianmarino/recommendation-system-approaches/blob/master/user-movie-genres-model/3.train-model-pandas-ray.ipynb)\n  * [3.2. Model training/validation (Spark version)](https://github.com/adrianmarino/recommendation-system-approaches/blob/master/user-movie-genres-model/3.train-model-spark.ipynb)\n\n\n## Requisites\n\n* [anaconda](https://www.anaconda.com/products/individual) / [miniconda](https://docs.conda.io/en/latest/miniconda.html)\n\n## Getting started\n\n**Step 1**: Clone repo.\n\n```bash\n$ git clone https://github.com/adrianmarino/recommendation-system-approaches.git\n$ cd recommendation-system-approaches\n```\n\n**Step 2**: Create environment.\n\n```bash\n$ conda env create -f environment.yml\n```\n\n**Step 3**: Enable project environment.\n\n```bash\n$ conda activate recommendations\n```\n\n## Open notebooks locally\n\n**Step 1**: Enable project environment.\n\n```bash\n$ conda activate recommendations\n```\n\n**Step 2**: Under project directory boot jupyter lab.\n\n```bash\n$ jupyter lab\n\nJupyter Notebook 6.1.4 is running at:\nhttp://localhost:8888/?token=45efe99607fa6......\n```\n\n**Step 3**: Go to http://localhost:8888.... as indicated in the shell output.\n\n\n**Note**: Use [Pycharm community](https://www.jetbrains.com/pycharm/download/#section=linux) to edit source code for more comfort.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fadrianmarino%2Frecommendation-system-approaches","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fadrianmarino%2Frecommendation-system-approaches","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fadrianmarino%2Frecommendation-system-approaches/lists"}