|
|
About the Scatter Plot
The Scatter plot in the Point Trending report allows you to see and estimate how one data point depends on another. For instance, one of the most effective uses of the Scatter Plot is to see how consumption is dependent upon Outdoor Air Temperature. By determining the dependencies like this, you can come to a better understanding of how to determine future energy needs based upon past performance.
This section will attempt to explain the use of the Scatter Plot for this scenario. That is, the dependence of Consumption on Outside Air Temperature.
The software will plot, or graph, on the X and Y axis all of the intersection points of Consumption (kWh) and Outside Air Temperature (T ˚F).
For instance, at any given date/time sample of kWh there is a corresponding date/time sample of Temperature. With T on the X-axis and kWh on the Y-axis, when you graph the two values, they represent one spot on the graph. The scatterplot software will plot all the values for T and kWh on the graph for the date range and rollup interval selected.
Figure 3-48 Example scatter plot
Then, the software attempts to determine the linear regression of the plotted points on the graph. The software will draw a straight line on the plotted points where the average error away from the line is the least amount possible for all points plotted. Lets say you could place a straight line on the plot and pivot the line at the center of all the plotted points. As you pivot the line around and move the line up and down, there will be a place where the average deviation (error) away from the line for all the points is the very least it could be. This is the place where the line stays. This is then the indication of linear regression.
Now in reality, a correlation coefficient of 1.0 would probably never be reached, but if the correlation coefficient is above 0.5, the software can draw the regression line in the scatter plot to help you better visualize the relationship. A correlation coefficient (r) of 0.0 would indicate that there is no dependency of the one data point to the other data point.
In general, squared r shows the part of Y defined by X, so for the plot example above where r=.7712, we can say that OutSide Air Temperature variation (X) is accounted for about 59% of Consumption variation (Y). (.77*.77=0.5929)
Once the line of regression is calculated, it is an easy exercise to calculate consumption for any given temperature. To calculate consumption for our example data points (kWh and Temperature), the software will use the simple formula kWh=A * Temp + B (y=A * x + B). A is the slope of the regression line, and B is the measurement or value at which the regression line crosses the 0˚F temperature.
Calculated consumption will not exactly match the observed consumption curve because there are factors other than temperature that affect consumption. This type of calculation is used in the Normalization for Degree Day calculations that the software can perform.
|
|
|
|