Score Algorithm Output

score_algorithm_output is a scoring function definition used in function exec_algorithm.

score_algorithm_output(algo_output, obs_ctrl)

Arguments

algo_output

(data frame) The output of exec_algorithm.

obs_ctrl

(list) (See break_signal)

Value

A data.table with the following columns:

k

The vector of \(k\) values

d_Idev

\(\Delta{\text{Idev}}\)

d2_Idev

\(\Delta^{2}{\text{Idev}}\)

k_score

The break score \(\to\) information deviation equation/model

k_sz

The number of observations in each \(k\)

geo_pmf.max

The maximal value of the geometric probability mass function returned by geo_pmf

Idev_wmean

The weighted-mean of \(\text{Idev}\) using as weights the geometric PMF (geo_pmf) in order to find the optimal \(k\)

d_kscore, d_Idev_wmean

For each \(k\), \(\Delta{k_\text{score}}\) and \(\Delta{\text{Idev}_\text{wmean}}\), respectively

tot_score

The mutual relative proportionality of curvatures over \(k\) and \(\text{Idev}_\text{wmean}\). The relative proportionality of \(\Delta^{2}{\text{Idev}_\text{wmean}}\) is subtracted from 1 to invert the proportionality. This makes tot_score the product of the minimum of one curvature and the maximum of the othe (gradient descent is not used to determine the optimal \(k\) because \(k\) is a discrete value.)

best_k

The value of \(k\) corresponding to the maximal value of tot_score

alt_k

The next best \(k\)

See also

Other Signal Processor Functions: exec_algorithm(), geo_pmf(), info_encoder(), signal_processor()