Overview: Regression Analysis
A regression analysis allows users to determine the statistical significance of a set of one or more measured external data sources relative to a single internal data source. The external data source is referred to as a predictor and the internal data source being influenced is referred to as the response.
Running regression analyses provides resulting values of:
-
Confidence, also referred to as the R2 value (adjusted R2 when using more than one predictor), determines the statistical 'fit' or relationship between the selected predictor(s) and the response. The confidence is a factor (value less than one). A good statistical fit value normally is above .90. A negative confidence value indicates that as the predictor values increase, the response decreases.
-
Constant, is the Y intercept. This value is used in creating a forecast model for predicting response data values
-
Coefficients, unique for each predictor, are values used in creating a forecast model for predicting response data values.
Why run a regression analysis?
Regression analysis is useful in determining the influence that a single activity or a combination of multiple activities has on your energy information. These influences include facility activities, internal processes, and weather conditions. You can select more than one Predictor Source because more than one such event can impact energy demand and usage.
Predictor equations can also be produced to determine future energy use, demand and cost characteristics, under similar conditions. In addition, EnergyCAP Enterprise can determine when, and how far, a data point is out of the ordinary for a typical day.
With an acceptable resulting confidence, the constant and coefficients are applied using the formula:
Predicted Response =
C1*Predictor1 Value + C2*Predictor2 Value + Cx*Predictorx Value
