The AMR tool is a customizable, semi-automated utility built into R to train and evaluate single or multiple machine learning models. It has features such as parameter sweeping to find the model that fits the desired metrics.
Once the report is generated, it contains the following information:
- A model description
- A model evaluation and comparison
- A ranking of the features
In general, the information described in the report can help increase the following:
- The quality of the feature variables
- The level of difficulty for the machine learning task
- Guidance for subsequent feature engineering and modeling
The AMAR tool is available at the following GitHub location: https://github.com/Azure/Azure-TDSP-Utilities/tree/master/DataScienceUtilities/Modeling ...