reduce_best_timecourse_params: Reduce to best timecourse parameters

In calico/impulse: Fit Impulse and Sigmoidal Curves to Longitudinal Data Using TensorFlow

Reduce to best timecourse parameters

Description

Across multiple fits of a timecourse summarize the best fitting timecourse in terms of least-squares error as well as by lowest absolute V within a tolerance of the least-squares set.

Usage

reduce_best_timecourse_params(
  timecourse_list,
  reduction_type = "loss-min",
  sufficiency_tolerance = 0.05
)

Arguments

timecourse_list	List output from estimate_timecourse_params_tf
reduction_type	How to choose the best parameter set, options are: loss-min: lowest loss function, loss-small-v-small: loss within sufficiency_tolerance of minimum loss and then minimize absolute sum of v_{inter} and v_{final} (useful primarily when not using priors).
sufficiency_tolerance	All timecourses within 1 + sufficiency_tolerance best fitting parameter set are deemed sufficient

Value

a list containing top parameter set and losses

calico/impulse documentation built on June 4, 2024, 5:28 a.m.