GraniteAI is an open source automated machine learning and data mining platform, usable in python or as a standalone application
The GraniteAI plotting library allows user to quickly visualise trends in their data, including getting statistics from within separate groups.
GraniteAI uses Bayesian optimisation to efficiently fit model hyperparameters, by trialling parameters that give the best improvement in accuracy per second. All models are built using scikit-learn pipelines that automatically preprocess the data, allowing numeric, categoric, and text data types, allowing a wide range of different inputs.
Understand the decisions being made using feature importance, showing the most important variables underlying the data, or reason codes, to see what impacts the model predictions the most.
The user interface, built in Qt, is designed to make powerful machine learning and data science more accessible. Users can quickly select variables to explore the data sets. With just a few clicks, users can start a new machine learning project by selecting a target variable, choosing features to learn from, and automatically training the models. After training, models can be used to make predictions, or get insights into the data.
