Obtain Coefficients from Scaled SVD of Census School Attendance Data

Obtain coefficients from a scaled SVD of school attendance data. The attendance data is derived from population censuses by the United Nations Statistics Division. The data is available in the "Population 5 to 24 years of age by school attendance, sex and urban/rural residence" table in from the Population Censuses' Datasets. The coefficients are a scaled version of the \(U\) matrix from the SVD.

Usage

coef_csa(data, n_comp = 5, eps = 1e-05)

Arguments

data

Data frame with attendance data. (Data from multiple files combined, but otherwise unprocessed).

n_comp

Number of SVD components to include in result. Default is 5.

eps

Parameter for truncating probabililities. Default is 0.00001.

Value

A tibble

Truncation and transformation

Reported proportions outside (eps, 1-eps) are shifted to eps or 1-eps, and all values are logit-transformed, before the SVD is applied.

See also

• data_ssvd_csa() Put school attendance data into the format required by the ssvd() function in bage

• tidy_csa() Format school attendance data into a tidy data frame.

Examples

coef_csa(un_csa_subset)
#> # A tibble: 150 × 5
#>    sex    country    time component     coef
#>    <chr>  <chr>     <int> <chr>        <dbl>
#>  1 Female Australia  2001 Component 1 -1.07 
#>  2 Female Australia  2006 Component 1 -1.10 
#>  3 Female Australia  2011 Component 1 -1.21 
#>  4 Female Australia  2021 Component 1 -1.23 
#>  5 Female Bulgaria   2001 Component 1  0.931
#>  6 Female Bulgaria   2011 Component 1  0.903
#>  7 Female Bulgaria   2021 Component 1  0.459
#>  8 Female Mexico     2000 Component 1  0.839
#>  9 Female Mexico     2005 Component 1  0.710
#> 10 Female Mexico     2020 Component 1  0.761
#> # ℹ 140 more rows