Just Another Machine Learner (JAML)

TLDR

JAML is a simple Python-based framework supporting full cycle of QSAR model development.

List of supported Machine Learning methods

• Naive Bayes classifier
• Bayesian regression
• Random Forest classifier
• AdaBoost
• k-NN classifier
• SVM classifier
• Deep Learning

JAML workflow

JAML workflow consists of the following steps:

• File submission
• Dataset creation
• Model training
• Prediction

Datamodel objects

• Files - submitted as SDF or CSV.
• Datasets - created from files by assigning semantic columns values (e.g. record id, continuous value, etc). All structures in datasets are standardized using one of the chosen standardizers.
• Models - trained from datasets by selecting the field (binary or continuous), descriptors and ML method(s).
• Resultsets (predictions) - similar to datasets, but result in predictions are attached as additional columns.

Running

API

python app.py

UI

For UI configuration and running please see UI README