Share
Explore BrainMass

Regression Model to Estimate Electrical Loads

Dear OTA â?"

I need a solution or method to estimate the hourly electrical load for the State of Georgia during 1994, using Regression Modeling Techniques.

Given will be Excel tables that provide Georgiaâ??s actual hourly weather and average electrical load for the years 1992 and 1993 (9293w_ld) and Georgiaâ??s actual hourly weather data for the year 1994 (94wtr) to estimate the average electrical load from using a regression model.

How the data is represented in the columns and cells in these tables can (and probably should) be revised (i.e., added, deleted, modified, rearranged, etc.) to make it easier to produce a viable model that outputs estimates for the year 1994 that closely shadow or mirror the trend or graph of the combined 1992 and 1993 actual average hourly electrical loads.

The Excel data files and details are attached, and if you have any questions please don't hesitate to ask.

Attachments

Solution Preview

Please see the explanation below as well as the two Excel sheets. I did see your note and filled in some data for the missing points.

-------------------------------------------

In this problem, we have two years of hourly weather and electrical load data, and want to set up a regression model which can predict the following year's electrical loads given similar weather data.

As a first step (in the file "9293w"), I performed a simple linear regression in Excel. This used all seven input fields and used all data without modification to predict load. The results were encouraging, showing a reasonable r-sqaure of 36% and a standard error of 484.

This model clearly had predictive power and makes intuitive sense, given that weather obviously influences electricity use and our large set of hourly data. But in a linear regressional model, it's important to avoid overfitting and to ensure that numerical inputs have truly numerical ...

Solution Summary

Regression models to estimate electrical loads are examined.

$2.19