I would like to model intraday electricity load data. I have quarter-hourly data for each day of the year.

In many countries daylight saving time is practive. This has the consequence that once a year a day is 1 hour longer (it has 25 hours) and once a year a day is shorter (only 23 hours). I don't think that changing the time (e.g. to UTC) is a solution here because people go to work at 8:00 a.m. in their "local" time not in UTC.

What I have found in the literature is (e.g. here MODELING AND FORECASTING SHORT-TERMELECTRICITY LOAD USING REGRESSION ANALYSIS). There the extra hour is discarded and the missing hours is filled with average values.

What is the most used procedure here? How do you cope with this probem?

I would like to model intraday electricity load data. I have quarter-hourly data for each day of the year.

In many countries daylight saving time is practive. This has the consequence that once a year a day is 1 hour longer (it has 25 hours) and once a year a day is shorter (only 23 hours). I don't think that changing the time (e.g. to UTC) is a solution here because people go to work at 8:00 a.m. in their "local" time not in UTC.

What I have found in the literature is e.g. here MODELING AND FORECASTING SHORT-TERMELECTRICITY LOAD USING REGRESSION ANALYSIS. There the extra hour is discarded and the missing hours is filled with average values.

What is the most used procedure here? How do you cope with this probem?