Fig. 3: Feature importance plots for composite severity prediction analyses. | npj Mental Health Research

Fig. 3: Feature importance plots for composite severity prediction analyses.

From: Predicting substance use behaviors with machine learning using small sets of judgment and contextual variables

Fig. 3

Balanced Random Forest (BRF) analyses were used to classify high versus low for the composite GAIN-SS score. The predictors are arranged according to the mean decrease in Gini scores with the most important predictors for classification listed at the top. Plots a–e correspond to BRF analyses with thresholds (th) of 1–5, respectively.

Back to article page