About Regression Aggregation Measuring Components

Regression aggregation measuring components are Aggregation Measuring Components used to represent Device Population groups. Unlike standard aggregation measuring components, these do NOT store aggregated measurement data. Instead, they are used to orchestrate M&V processing for their associated Device Population groups and their corresponding Control Groups (see Understanding Device Population and Control Groups in the Administrative Use Guide for more information about the relationship between Device Population and Control groups).

Regression aggregation measuring components are created via the Dimension Scanning process based on changes to Devce Populaion groups.

Regression aggregation measuring components are defined by the following:

• Initiating Measuring Component Set: The measuring component set that triggered creation of the aggregation measuring component via the Dimension Scanning process (see Dimension Scanning for more information).

• Status: The current status of the measuring component. Valid values are "Active", "Aggregate", and "Inactive".
• Service Type: The type of service (electric, water, and so on) the measuring component supports
• Time Zone: The time zone in which the measuring component operates

• Processing Information: Details of aggregation processing for the aggregation measuring component, including the date and time through which aggregation has been performed.

• Measuring Component Attributes: Attributes that describe the dimensions and criteria for the aggregation measuring component (used by the Dimension Scanning process)

System Data

Regression aggregation measuring components are based on the following system data:

Data Type	Description
Maintenance Object	Measuring Component (D1-MEASRCOMP))
Business Object(s)	Regression Aggregator for M&V (X1-RegressionAggregatorMV)

Regression Aggregation Measuring Component Processing

As noted above, regression aggregation measuring components orchestrate M&V processing for their associated Device Population groups and their corresponding Control Groups.

Over time, as program events take place, Event Participation records record which customers participate in which events and the outcome of those events (such as "Successful", "Error", and so on). These are used to create Service Point Event Outcome Staging records.

On a regular basis, the Aggregation by MC Set Monitor (D1-AMCS) batch process triggers processing for aggregation measuring components in the "Active" state. This initiates a series of algorithms which do the following:

• Check for participation in program events
• Build a list of customers from the Device Population group who successfully participated in the event (Find Constituents) in the Entity Group Constituents Temporary Table (D1_GTT11).
• Trim the list of constituents by retrieving related Service Point Event Outcome records and deleting records from the Entity Group Constituents Temporary Table (D1_GTT11) in the event of any of the following:
  • The record indicated an outcome of non-responsive device, or canceled
  • The record has the Multiple Entity Group flag set to "Yes"
  • Any customer where another thermostat for the service point has participated in a different event for the same external event ID.
  • If no Service Point Event Outcome record is found for a customer in the Device Population group
• Perform a "Difference in Differences" regression analysis, aggregating and comparing data from the Device Population group to corresponding data from the Control group, and create a M&V transaction, which uses M&V Calculation Rules to perform calculations based on the regression outcome quantities
• Check for participation in any remaining unprocessed qualifying events
  • If no event occurred, the measuring component remains in the "Active" state
  • If another event is found, the measuring component is set to the "Aggregate" state (and can be transitioned back to the "Active" state via the Transition to Default Next Status (Java) (F1-AT-RQJ) batch process.