https://github.com/cthoyt/yamc

Yet another ML method comparison

https://github.com/cthoyt/yamc

Science Score: 23.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • CITATION.cff file
  • codemeta.json file
  • .zenodo.json file
  • DOI references
    Found 2 DOI reference(s) in README
  • Academic publication links
    Links to: acs.org
  • Academic email domains
  • Institutional organization owner
  • JOSS paper metadata
  • Scientific vocabulary similarity
    Low similarity (2.6%) to scientific vocabulary
Last synced: 10 months ago · JSON representation

Repository

Yet another ML method comparison

Basic Info
  • Host: GitHub
  • Owner: cthoyt
  • License: mit
  • Default Branch: main
  • Size: 3.85 MB
Statistics
  • Stars: 0
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Fork of PatWalters/yamc
Created over 3 years ago · Last pushed almost 4 years ago

https://github.com/cthoyt/yamc/blob/main/

# yamc
Yet another ML method comparison, a comparison of 
* Gaussian Process Regression (GP)
* XGBoost (XGB)
* FeedForward Neural Network (FFNN)
* ChemProp (CP)

We are only comparing algorithms for GP, XGB and FFNN. The same descriptors (RDKit Morgan2) are used for all three. 

Datasets are from [https://pubs.acs.org/doi/10.1021/acs.jcim.8b00542](https://pubs.acs.org/doi/10.1021/acs.jcim.8b00542)

## Procedure

1. fixed_comparison.py - runs GP, XGB, and FFNN, writes output to comparison.csv
2. run_chemprop.py - runs ChemProp (this takes several days, I could probably make it faster)
3. analyze_chemprop.py - gathers ouptut from ChemProp and writes output to cp_comparison.csv
4. analysis.ipynb - reads comparison.csv and cp_comparison.csv and generates plots

## Random Splits 

![](Random_Split_r2.png)

![](Random_Split_rmse.png)

## Scaffold Splits 

![](Scaffold_Split_r2.png)

![](Scaffold_Split_rmse.png)

Owner

  • Name: Charles Tapley Hoyt
  • Login: cthoyt
  • Kind: user
  • Location: Bonn, Germany
  • Company: RWTH Aachen University

GitHub Events

Total
Last Year