Table 4 Model validation metrics on the five different trained data.

From: Non-parametric quantile regression-based modelling of additive effects to solar irradiation in Southern Africa

Location

Model

BG test

Box Ljung

R-square

\(80\%\)CV

\(20\%\)CV

p-value

p-value

Windhoek

PLAQR

0.117

0.501

0.929

0.963

0.964

AQR

0.247

0.876

0.910

0.953

0.954

QGAM

0.147

0.858

0.934

0.966

0.967

Alice

PLAQR

0.329

0.204

0.923

0.962

0.961

AQR

0.631

0.616

0.913

0.957

0.956

QGAM

0.631

0.709

0.930

0.966

0.964

Durban

PLAQR

0.998

0.664

0.920

0.957

0.959

AQR

0.889

0.436

0.911

0.952

0.955

QGAM

0.889

0.208

0.924

0.960

0.962

Pretoria

PLAQR

0.243

0.078

0.929

0.962

0.964

AQR

0.776

0.656

0.917

0.955

0.957

QGAM

0.776

0.148

0.935

0.966

0.967

Venda

PLAQR

0.939

0.994

0.914

0.960

0.956

AQR

0.806

0.921

0.907

0.955

0.952

QGAM

0.806

0.058

0.923

0.965

0.961