The teller is a model-agnostic tool for Machine Learning (ML) explainability. It uses Taylor series and finite differences to explain ML models predictions:

a little increase in model’s explanatory variables + a little decrease = approximate sensitivities of its predictions to changes in these explanatory variables

The teller is now available on Pypi (yeaaah!), and can be installed from the command line as:

pip install the-teller

The code is also documented on readthedocs (it’s a work in progress):

https://the-teller.readthedocs.io/en/latest/?badge=latest

For those who haven’t had a taste of the teller yet, these notebooks will constitute a good (and fun) introduction:

• Heterogeneity of marginal effects
• Significance of marginal effects
• Model comparison
• Classification
• Interactions

Note: I am currently looking for a gig. You can hire me on Malt or send me an email: thierry dot moudiki at pm dot me. I can do descriptive statistics, data preparation, feature engineering, model calibration, training and validation, and model outputs’ interpretation. I am fluent in Python, R, SQL, Microsoft Excel, Visual Basic (among others) and French. My résumé? Here!