Application of Machine Learning Approaches for Classifying Sitting Posture Based on Force and Acceleration Sensors

Figure 2

Classification accuracy of the seven different sitting positions (as shown in Figure A2 (Supplementary Materials); 1: upright position, 2: reclined position, 3: forward inclined position, 4/5: laterally tilted right/left position, and 6/7: crossed legs, the left leg over the right one/the right leg over the left one) for the three best performing classification methods (NN (beige), Boosting (green), and RF algorithm (red)), each using the parameters leading to the highest classification accuracy. The horizontal lines represent the classification accuracy of Tan et al. [19] and Mota and Picard [21] (black) as well as that of the current study using RF (red).