I am dealing with classification predictive models in the context of machine learning; I am using different models (KNN, SVM, Random Forest, Logistic regression) and I am using the function varImp from package caret to extract feature importance.
Is it correct to say that since the most important features are the most useful for classification, then the most important features are possible "risk factors" for the target variable?
$\begingroup$
$\endgroup$
1
-
$\begingroup$ This is ambiguous and unnecessary language, variables with high scores have more impact on your target, that is changing these variables has a bigger impact on the prediction. Keep it simple. $\endgroup$– user2974951Commented Feb 3, 2023 at 15:26
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
1
The term 'Risk factor' is vague, its not clear whether you're referring to a modifiable factor that if we were to alter would affect your outcome, or a predictive factor whose value is correlated with risk. If the latter then its fine, for the former you need more information from subject matter theory or causal analysis methods.
-
$\begingroup$ With the term "risk factor" I am referring to a variable that is statistically associated with an increased (or decreased) risk of an event (e.g. a disease) $\endgroup$– autu_mnCommented Feb 3, 2023 at 14:48