Towards a Complete Set of Gym Exercises Detection Using Smartphone Sensors

Table 6

Table of classification results.

S number	Sensor name	Smartphone positions	Naïve Bayes classification accuracy in percentage (%)	K-NN classification accuracy in percentage (%)	Decision tree (J-48) classification accuracy in percentage (%)	Sensors features (mean, std. deviation, min, and max) A (x, y, z) G (x, y, z)
Single sensor used
1	Accelerometer	Arm	52.61	95.87	87.73	12
2		Belly	40.99	95.16	86.29	12
3		Leg	47.59	96.39	91.58	12
4	Gyroscope	Arm	31.93	91.83	80.23	12
5		Leg	23.27	87.39	75.00	12
Two sensors used
6	Accelerometers = 2	Arm and belly	69.92	98.40	92.96	24
7		Arm and leg	74.18	99.17	95.31	24
8		Belly and leg	63.10	98.65	93.85	24
9	Gyroscopes = 2	Arm and leg	42.52	96.29	86.40	24
10	Accelerometer + gyroscope	Arm
		62.49%	98.04	90.63	24
11		Leg	47.72	97.76	91.40	24
Three sensors used
12	Accelerometers = 2, gyroscope = 1	Arm and belly	77.42	99.41	93.94	36
13	Accelerometer = 2, gyroscope = 1	Belly and leg	77.59	99.02	94.21	36
14	Accelerometers = 3	Arm, belly, and leg	79.59	99.51	89.0	36
Four sensors used
15	Accelerometers = 2, gyroscopes = 2	Arm and leg	79.32	99.63	96.05	48
Five sensors used
16	Accelerometers = 3, gyroscopes = 2	Arm, belly, and leg	80.72	99.72	96.29	60