local_polynomial_fits: Fit local polynomial models to estimate derivatives
In rossklin/dynpan: Dynamics estimation for multiple time series

Applies local linear/polynomial regression around points and differentiates the results.

1 2	local_polynomial_fits(tt, points, k, timesteps = NULL, times = NULL, degree = 2, weight.function = constant_weight_function, vars = NULL)

`tt`	`time.table` contain time series controlled by some dynamics
`points`	points at which to estimate derivative
`timesteps`	timesteps around each neighbours to base estimate on
`times`	exact (relative) time values to base estimate on
`degree`	degree of local polynomial fit
`weight.function`	weight function to use in the local polynomial fit
`vars`	variables from tt and points to include in result (those from `tt` taking precedence)

At least one of timesteps or times has to be specified, if only the former is specified the value of times is computed using the time delta of tt.

degree should be low, since each estimate will in general be based on only a few data points.

rossklin/dynpan documentation built on May 27, 2019, 11:39 p.m.