How is the final part after Model building is presented to business?


After we build a logistic regression model , ie. even after model validations like Concordant-disconcordant pair, lift gain curve, ROC curve, AUC, Confusion matrix

.How it is presented to business?Generally how a model robustness is checked in a time period as well?

Could anyone please explain it with an example.


If I understand your question correctly, you mean how to evaluate a model in a real situation with newly collected data. If this is the case, we can name it as evaluating model generalization ability.
For example if we consider a fault detection model, after all sort of evaluations performed on the model and it showed satisfactory results, a period of observation - lets say 3 months - needs to be defined with a manual validation phase.
During this period the model and manual validation will be happen in parallel. At the end of the period, the model will be evaluated against the manual evaluation stage and will compare with previous performance metrics, if the model could maintain results compare to preliminary results then the model got the generalization ability otherwise model needs to gone trough a refining process and the test phase should repeat again.


Lift Chart and Rank Ordering is mostly used in industry.


It starts with a problem statement. It should end with a solution statement addressing the problem. Doesnt matter what chart, what graphs you produce. It should answer the problem and help your audience understand your solution providing a scope of real time implementation and way forward.


I agree with your point of keeping Problem Statement and what solution we are bringing on table for Business.

I just quote chart as Lift Chart is easiest chart to explain to Non-Data Scientist Person.

Anyway we as Data Scientist responsibility to convey the result to business in succinct and simplified way.