Table 2 The total number of hours falling into each category is shown.

From: Developing reliable hourly electricity demand data through screening and imputation

Screening Algorithm First Screening: 54 BAs Second Screening: 54 BAs
Counts Percent (%) Counts Percent (%)
‘okay’ 1,843,424 97.36 1,892,736 99.962
‘missing’ 41,015 2.16 0 0.000
‘negative or zero’ 863 0.05 0 0.000
‘identical run’ 3,861 0.20 608 0.032
‘global demand’ 170 0.01 0 0.000
‘global demand ± 1 hour’ 285 0.01 0 0.000
‘local demand’ 671 0.03 23 0.001
‘double-sided delta’ 1,212 0.06 28 0.001
‘single-sided delta’ 181 0.01 15 0.001
‘anomalous region’ 1,773 0.09 46 0.002
Flagged for Imputation 50,031 2.64 720 0.038
Flagged for Imputation: Ignore Identical Runs 46,170 2.50 112 0.006
Total Hours 1,893,456 100.00 1,893,456 100.00
  1. The second round of screening, after the initial screening and imputation, shows a vast reduction in the quantity of hours flagged as anomalous. This reduction indicates strong anomaly screening and imputation performance.