adapt_pic_with_sigma2maxmodel: Adapts the object that contains PIC for all candidate C's and...

In RCTS: Clustering Time Series While Resisting Outliers

Adapts the object that contains PIC for all candidate C's and all subsamples with sigma2_max_model.

Description

The PIC is calculated with a sigma2 specific to the configuration (= number of groups and factors). Because the method to estimate the number of groups and factors requires sigma2 to be equal over all configurations (see proofs of different papers of Ando/Bai) we replace sigma2 by the sigma2 of the configuration with maximum number of groups and factors (this is the last one that was executed).

Usage

adapt_pic_with_sigma2maxmodel(df, df_results, sigma2_max_model)

Arguments

df	contains PIC for all candidate C's and all subsamples
df_results	dataframe with results for each estimated configuration
sigma2_max_model	sigma2 of model with maximum number of groups and factors

Value

data.frame of same size as df

Examples

set.seed(1)
df_pic <- data.frame(matrix(rnorm(4 * 50), nrow = 4)) #4 configuration / 50 candidate values for C
df_results <- data.frame(sigma2 = rnorm(4))
pic_sigma2 <- 3.945505
adapt_pic_with_sigma2maxmodel(df_pic, df_results, pic_sigma2)

RCTS documentation built on May 31, 2023, 9:20 p.m.