Table 7.

Mean testing performance.

Class	Precision	Recall	F1-score	Accuracy	N(segments) in test set
Text features only
Hotspots	0.443	0.675	0.530		4
Non-hotspots	0.652	0.435	0.469		5
Weighted average/Total(N)	0.568	0.546	0.501	0.545	9
Speech features only
Hotspots	0.543	0.592	0.534		4
Non-hotspots	0.603	0.560	0.553		5
Weighted average/Total(N)	0.586	0.565	0.543	0.566	9
Multimodal (text and speech features)
Hotspots	0.464	0.617	0.525		4
Non-hotspots	0.594	0.495	0.512		5
Weighted average/Total(N)	0.543	0.556	0.522	0.555	9

Note. Per class and average performance scores for the final models.