. 2022 May 7;22(9):3555. doi: 10.3390/s22093555

Table 3.

Supervised Classification Models. Summary table of the compared models to perform the joint segmentation and classification of time points into walking and non-walking activities. For each model, hyper-parameters to be tuned are listed (following the tidymodels (https://parsnip.tidymodels.org (accessed on 17 March 2022)) naming conventions) along with the grid that was used for tuning. The notation x:y:z is a shorthand for all values between x and z included with a step of y.

Model	Hyper-Parameters	Tuning Grid
Decision Tree	`cost_complexity` `tree_depth`	$10^{- 10 : 1 : - 1}$ $1 : 1 : 10$
Radial Basis Function SVM	`cost` `rbf_sigma`	$2^{- 5 : 1 : 5}$ $10^{- 10 : 1 : 0}$
k-NN	`neighbors` `dist_power`	$1 : 2 : 67$ $1, 2$
Logistic Regression	`threshold`	$0.05 : 0.05 : 0.95$