Can adding more data, influence the SHAPley values? I am having a classification dataset. Everytime I add more data to the dataset the SHAP values are getting changed and the top list of features are changed.
Can anyone explain what is this behavior? Am I doing any mistake somewhere?
and F1-score also getting reduced on adding more data. Is it some kind of data quality issue?
 A: One possible explanation: it depends what you mean by adding data because:

*

*shapley value shows the individual contribution of a feature that is above or below the average contribution or in other words:


This is the predicted value for the data point x minus the average
predicted value

or in other words efficiency of a shap-value
https://christophm.github.io/interpretable-ml-book/shapley.html 5.9.3.

*

*further, features give the same amount of contribution to a model, if they also contribute equally to all sorts of coalitions, that means: shap values can fairly distribute both gains and costs to several actors (features) working in coalition
https://www.investopedia.com/terms/s/shapley-value.asp, and also 5.9.3 (also called symmetry)

So to answer your question: every amount of data you add, may change the symmetry of a features contribution overall, and thus its importance, and in some way the efficiency, as you probably press new never seen values into the model, that change the average prediction, and thus the contrubtuion of a feature against an average benchmark.
