Skip to main content

Table 3 Stepwise variable selection on dataset C

From: Predicting emergency department visits in a large teaching hospital

Variable

Coefficient

Standard error

Significance level

Intercept

308.6042

44.3327

<.01

Monday

12.5382

1.0441

<.01

Friday

9.718

1.0134

<.01

Saturday

− 8.5790

1.0223

<.01

Sunday

− 11.0413

1.0386

<.01

Winter

4.6875

1.0734

<.01

Time trend (daily)

0.0066

0.0013

<.01

Carnival

8.1048*

4.2351

N.S.

Holiday

− 14.0091

3.2413

<.01

Monday * Holiday

− 9.2721

4.7393

N.S.

Sunday * Holiday

13.9759

5.0697

<.01

Summer vacation (week 1 + 2)

− 12.1128

1.5751

<.01

Summer vacation (week 3 + 4)

− 16.4741

1.6734

<.01

Summer vacation (week 5 + 6)

− 6.4804

1.6781

<.01

Max temperature (in 0.1°C)

0.0196

0.0084

<.05

Global radiation (in J/cm2)

0.0038

0.0008

<.01

Max pressure (in 0.1 hPa)

− 0.0210

0.0043

<.01

Max visibility (in 100m)

− 0.0100

0.0044

<.05

Max humidity (in %)

− 0.1357

0.0726

N.S.

Snow (in hours)

− 0.7926

0.3286

N.S.

Fog (in hours)

0.4101

0.1583

<.01

Storm (in hours)

− 0.8429

0.4619

N.S.

Observations

921

Multiple R-squared

0.4858

Adjusted R-squared

0.4738

  1. Dataset C is based on calendar variables plus weather predictions at the Jeroen Bosch Hospital (the refined set of predictors containing the best predictors of datasets A and B); * = interaction; N.S. not significant