What is model interpretation?
As anybody can read the model, through the sensitivity analysis and the live predict, you can use these visuals to explain your model to your stakeholders. You are a marketing manager and want to explain to your customer why you want to put more budget into facebook? Take your live predict feature and make a slideshow out of it. It shows the dependency between the conversions and the Facebook ads.
What if the customer asks you how you got those results?
How can I interpret the results?
TADA provides a mathematical formula for everyone to read. Your customer can see how the dependency between facebook marketing budget and conversion is established. This formula is readable by anyone. It is reassuring because it shows how TADA generates its predictions. The results are not generated out of a complicated neural network that is not readable. Your customers can ‘read’ you model, so can you.
Proof of my model
Everybody can see with their own eyes where the model comes from. It is not a magical wand hidden in a black box which states unverifiable truths. It is completely open, transparent, accessible for everyone to see.
Read ‘deploy your model’ (mettre lien deploy your model) to see how this mathematical formula is used in production.