Table 2. Results for the fish size regression errors (in cm) of different regressors.

Best values are marked in bold.

Model	MAE	MSE	$R^2$	MAPE	Time (s)
Extra trees	1.8613	8.7115	0.7694	0.1173	0.101
CatBoost	1.8506	8.8161	0.7668	0.1172	1.211
Gradient boost	1.8504	9.3102	0.7544	0.1166	0.075
Random forest	1.8830	9.5934	0.7474	0.1175	0.201
Light GBM	1.9224	9.5624	0.7471	0.1201	0.021
XGBoost	1.9853	9.8369	0.7409	0.1252	0.076
$k$-NN	2.0806	10.1672	0.7312	0.1331	0.005
Linear	2.5980	15.2516	0.6071	0.1656	0.127
Ridge	2.5973	15.2517	0.6071	0.1655	0.003
Bayesian ridge	2.5962	15.2524	0.6071	0.1655	0.003
Least angle	2.6365	15.518	0.5993	0.1676	0.003
Huber	2.4311	16.4486	0.5823	0.1585	0.005
Decision tree	2.6236	17.0694	0.5577	0.1617	0.007
Lasso	2.7333	19.2231	0.5162	0.1769	0.004
Elastic net	2.7526	19.3541	0.5145	0.1768	0.003
OMP	2.6763	21.8864	0.4456	0.1753	0.003
AdaBoost	4.3506	29.9264	0.1917	0.3018	0.035
PAa	3.8979	31.7804	0.1534	0.2338	0.004
LLAb	4.3817	39.4312	−0.0028	0.2706	0.003
Dummy	4.3817	39.4312	−0.0028	0.2706	0.002

Notes:

The best 20 of a total of 25 are shown, ordered by ascending mean square error (MSE). Error is provided using several common metrics (MAE, MSE, $R^2$, MAPE). The total time (Time) in seconds (s) is also provided for comparison of regression performance.

^aPassive–aggressive.

^bLasso least angle.