Table 3.

Cross-validation of LWPR models. Models were learned using all runs of the same skier (columns). Mean errors and standard deviations of the COM positions of these models were calculated for all 18 runs. All values in the table are given in m.

		MODEL TRAINING / SKIER
SKIER	RUN	1	2	3	4	5
1	1	0.042 (0.019)	0.058 (0.018)	0.076 (0.030)	0.052 (0.021)	0.090 (0.041)
	2	0.032 (0.012)	0.046 (0.022)	0.057 (0.021)	0.030 (0.012)	0.066 (0.029)
	3	0.034 (0.012)	0.047 (0.022)	0.067 (0.031)	0.041 (0.018)	0.076 (0.035)
2	1	0.074 (0.031)	0.048 (0.020)	0.073 (0.027)	0.056 (0.027)	0.065 (0.035)
	2	0.059 (0.026)	0.040 (0.014)	0.066 (0.018)	0.053 (0.020)	0.076 (0.020)
	3	0.052 (0.021)	0.031 (0.013)	0.058 (0.026)	0.041 (0.020)	0.063 (0.027)
	4	0.054 (0.020)	0.041 (0.013)	0.051 (0.013)	0.054 (0.020)	0.064 (0.019)
3	1	0.063 (0.025)	0.050 (0.026)	0.042 (0.021)	0.061 (0.023)	0.046 (0.014)
	2	0.050 (0.022)	0.049 (0.021)	0.042 (0.018)	0.053 (0.025)	0.050 (0.028)
	3	0.077 (0.015)	0.077 (0.020)	0.049 (0.016)	0.076 (0.025)	0.069 (0.016)
	4	0.056 (0.024)	0.057 (0.019)	0.040 (0.021)	0.057 (0.022)	0.051 (0.025)
4	1	0.055 (0.022)	0.048 (0.021)	0.056 (0.021)	0.054 (0.013)	0.062 (0.026)
	2	0.047 (0.017)	0.049 (0.029)	0.063 (0.019)	0.026 (0.015)	0.058 (0.021)
	3	0.061 (0.023)	0.059 (0.023)	0.076 (0.025)	0.034 (0.013)	0.069 (0.025)
5	1	0.071 (0.028)	0.055 (0.025)	0.063 (0.026)	0.062 (0.029)	0.049 (0.021)
	2	0.083 (0.032)	0.063 (0.026)	0.066 (0.033)	0.066 (0.031)	0.045 (0.022)
	3	0.063 (0.025)	0.064 (0.020)	0.049 (0.018)	0.058 (0.028)	0.043 (0.017)
	4	0.060 (0.025)	0.056(0.015)	0.047 (0.018)	0.055 (0.022)	0.039 (0.017)
method : LWPR		COM MEAN ERRORS (STANDARD DEVIATIONS)