Radius_mean, perimeter_mean, area_mean, concavity_mean, concave_points_mean are correlated with diagnosis so we could use them to create and test the model.

Radius_mean, perimeter_mean, area_mean are highly correlated with each other, we can use one of them or all of them to create the model.

Only radius_se and perimeter_se are just slightly correlated with diagnosis so we will not use the SE features to create the model.

Radius_worst, perimeter_worst, area_worst, concavity_worst, concave-points_worst are correlated with diagnosis so we will use them to create the model.

Radius_worst, perimeter_worst, area_worst are highly correlated with each other we can use one of them or all of them to create the model.