sronilsson
commented
9 months ago Hoi @mrnels19 !
The .meta.csv file is a SimBA model config file. It also functions as a log for you to see which settings you used to create your classifier. You can use it as a config file in the SimBA interface to load the settings, in cases where you want to create a new model with the same or similar settings here:
The feature_importance_bar_graph and feature_importance_log csv file stores a list the most important features, and the importance of all features, within your classifier. For example, if you where classifying running, you'd expect features that measures movement and velocity to be shown at the top. These two files get created when you have these checkboxes ticked when you create your model:
They don't give you the accuracy of the model, you can think of it has information on how the model works and reaches it conclusions and sanity check.
You can read more about these settings in the documentation HERE.
Hi, I have been using SimBA to train a few models, and I was wondering what the documents in the model_evaluations folder represent. In this folder, I have a feature_importance_bar_graph saved as a .png file, a feature_importance_log csv file, and a meta csv file. What information is found in each of these files, and how does this relate to the accuracy of the model? Thank you!