Skip to main content

A standalone module to build a recommender pipeline

Project description

recommendation_engine

Repo for the recommendation engine that was part of the DRP project

Recommender Pipeline

Steps to implement a recommender pipeline (Specific implementation of this pipeline is available in ./recommender/recommender_pipeline.py)

  1. Generate reaction features

    • Get the chemicals in a reaction. For DRP these are referred to as triples
    • Generate descriptors for each of the chemicals in the reaction
    • Generate a sampling grid of reaction parameters
    • Expand grid by associating descriptors with each point on the grid
  2. Run trained models with the reaction Sieve

    • Get a trained machine learning model
    • Filter sampling grid by running it through the ML model
    • Make a list of all the potentially successful reactions as predicted by the ML model
  3. Recommend reactions

    • Calculate the mutual information of the potential reactions as compared to the already completed reactions
    • Select the top 'k' reactions with the highest MI

Progress

  • Generate Reaction features
  • Reaction Sieve
  • Reaction Recommender
  • Test and evaluate against Nature paper

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

chemrecommender-0.0.1.macosx-10.7-x86_64.tar.gz (13.7 kB view hashes)

Uploaded source

Built Distribution

chemrecommender-0.0.1-py3-none-any.whl (8.7 kB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor NVIDIA NVIDIA PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page