brandLogo: alt="EnergyLogo"

About the Correlation Report

---

The Correlation Report allows you to correlate two logs of data over a user defined period of time. This can be useful for determining the relationship between any two data points. It allows the user to select the data points to use for the x and y axis of a correlation chart (scatter plot), and then use the configuration parameters to enable a linear regression trend line (automatic or custom), set up linear regression forecasting, and tweak the tabular results. The Correlation Report also leverages from the other common features available to all E2 Profiler reports.

The Correlation report can be useful for determining a relationship between energy data logs and timestamped unit production logs. This can allow you to forecast energy usage based on predicted production units.

Figure 4-20 Correlation Report Example.

This example shows a correlation run for the electric demand of a site, and its corresponding units of production (widgets) over a month time period. You will notice that there is a strong correlation between the two data points (i.e. the correlation coefficient, r, is close to 1). The forecasting features allow the linear regression trend line to be extended in order to predict electric demand based on its historical association to widgets produced. The table below the chart gives a breakdown of the linear regression (the table granularity is also user-definable).

Creating a Correlation Report.

To create a Correlation Report, perform the following tasks:

• Start the Correlation Report
• Select the Data Points to Correlate
• Set Report parameters

Start the Correlation Report

1. Launch shortBrandName from your browser using the following URL: http://hostname/eas
   where hostname is the name or IP address of the Web Supervisor.
2. Enter appropriate login information to gain access to the home page.
3. From the home page, click on the Correlation Report link to open the Correlation Report.

Select the Data Points to Correlate

1. Expand the tree to find the site and data point(s) you wish to chart. Right click on a data point and select Add to Report.

   The data point appears in the Selection window.

   Note: If you select a site or a meter and do an Add to Report, the software will aggregate all data points of the same type that are subordinate to that selection.

2. Repeat adding a data point or copying and pasting data points until you have at least two data points in the report that you want to correlate.

Set Report parameters

Figure 4-21 Correlation Report parameters

1. In the Period frame, select the time period for the data and select the days of the week to exclude from the report (if any).
2. Set Rollup Interval in days, hours, and minutes.

The interval defines a block of time within which the data samples will be reported. Depending on the type of data, the data samples are handled differently. For demand data, the highest value sample, or peak, collected within the interval will be used. For consumption data, all the samples within the interval will be added together. For all other data types, the data samples are averaged together.

3. In the Normalization frame, set your desired parameters.

If you want to normalize for Floor area, for weather, or production, refer to "Types of Normalization", for details.

4. Adjust the following parameters, as desired:
5. Correlation

   The scatter plot chart allows you to see and estimate how two data points depend upon each other. A correlation coefficient (r) can go from 0, meaning no dependency, to 1, meaning a direct correlation and dependence between the two points. You can define which point is on the X axis and which point is on the Y axis (these selections will automatically be populated when the first two data points are added to the report, but you are free to change as needed). You can also lag the scatter plot of Y for any period of days, hours, and minutes.

6. Lag

   Sometimes there is a time lag between cause and effect: for example, it takes some time for the building system to react to a temperature change. In this case we can make dependence more pronounced if we would use temperature measured some time prior to measurement of the consumption value. This time shift is a lag. By varying lag we can get the highest possible correlation between two variables.

Figure 4-22 Plotted values for a 30 minute time lag setting

So you can use the Lag setting to make the software adjust the compared data points by shifting correlated values in degrees of Days, Hours, or Minutes.

• Linear Regression

  You can use the linear regression feature to draw a trend line with your data. This is useful for forecasting. The default selection is "auto" which means that the linear regression (best fit) will automatically be calculated for the data. You can optionally disable the linear regression feature ("none" selection) or enter your own custom linear regression ("custom" selection). If you enter your own "custom" linear regression line, you can use the Slope and Offset fields to enter the parameters for the trend line. In "auto" mode, the slope and offset will automatically be calculated for the line of best fit.

• Forecasting

  Sometimes there is a time lag between cause and effect: for example, it takes some time for the building system to react to a temperature change. In this case we can make dependence more pronounced if we would use temperature measured some time prior to measurement of the consumption value. This time shift is a lag. By varying lag we can get the highest possible correlation between two variables.

• Table Granularity

  If you have the linear regression feature enabled (i.e. either “auto” or “custom” mode), you can use the forecasting feature to extend the linear regression, thus allowing forecasting. You can select to forecast either the x or y axis, and then you can enable forecasting by selecting the "On" checkbox and entering the forecast value. When the report is run, the linear regression will be extended to the forecasted value for either the x or y axis, depending on your selection.

• Click on Run Report to create the Correlation Report.

  The Correlation Report is displayed, as shown in Figure 4-23.

Figure 4-23 Correlation Report