Table 6.

Regression accuracies of shallow machine learning algorithms to predict the mixing time remaining of the honey-water blending process. E—Energy, SAA—Sum Absolute Amplitude, G—Gradients of Features, PCs—Principle Components, DWT—Discrete Wavelet Transform.

	ANN (R2)			SVM (R2)			LSTM (R2)
Features	Non-central	Central	Combined	Non-central	Central	Combined	Non-central	Central	Combined
E, SAA	0.751	0.788	0.920	0.276	0.223	0.795	0.852	0.818	0.954
E, SAA, G	0.875	0.806	0.922	0.894	0.663	0.780	0.755	0.853	0.969
PCs	0.810	0.882	0.956	0.622	0.503	0.817	0.705	0.584	0.959
PCs, G	0.910	0.904	0.972	0.771	0.190	0.721	0.939	0.936	0.969
DWT, E, SAA	0.862	0.960	0.949	0.713	0.570	0.907	0.963	0.895	0.957
DWT, E, SAA, G	0.758	0.892	0.948	0.827	0.722	0.899	0.965	0.865	0.959
DWT, PCs	0.786	0.881	0.957	0.552	0.551	0.806	0.857	0.751	0.973
DWT, PCs, G	0.916	0.914	0.972	0.787	0.503	0.854	0.914	0.930	0.972