Use class weights for imbalanced data (link):
Each algorithm has a standard score which will be taken automatically. For example logistic regression takes accuracy. You can easily look that up.
With GridSearchCV you can easily find the best parameters (link). MOre about sample weight can be found in this video:
Darstellung als Dataframe:
Wir wollen weitere Metriken (precission & recall) ebenfalls über GridSearchCV erhalten. Außerdem wollen wir die Ergebnisse für die Test- und Training-Daten ausgegeben bekommen (link).
Plot results for test set and train set (link). They might look very different!!!:
Add a custom metrics to your GridSearchCV. For example a metric which balances Racell and Precession (link and link and link) —> make_scorer:
