The alert is meant to prompt the user to think about and identify what could be causing the high usage anomaly and take any action, if it is within their control.
Bidgely High Usage Alerts
The first action item when receiving the high usage alert should be to think back on whether there may have been any energy consumption anomalies in the past week (one of the days likely to have had high unusual usage). If such an anomaly is identified by the user, subsequent actions will depend on the user’s motivations.
- If the user wants to counteract the high usage day to stabilize their ultimate bill, they can take particular care with energy saving actions for the rest of the billing cycle to mitigate the impacts of high energy consumption. Actions may include turning down the temperature to save on heating, waiting longer before running washer/dryer loads, etc. A user may also be directed to any utility programs for energy savings to help save.
- If a user does not care to save energy for that billing cycle, no action needs to be taken at the time.
Currently, the high usage alert does not specifically state the day which it is referring to. However, it is usually triggered within a 1-2 day period if we are receiving timely data from the utility. If there are delays in utility data ingestion, then the high usage event will be delayed accordingly. Generally, the user can assume that the email is regarding a day within the past week.
The high usage alert has 4 different variations:
- High usage compared to an average weekday within the billing cycle
- High usage compared to an average weekend within the billing cycle
- High usage compared to average weekday peak hour consumption within a billing cycle (for electric users only)
- High usage compared to average weekend peak hour consumption within a billing cycle (for electric users only)
- As defined, weekend refers to both Saturday & Sunday and weekday refers to Monday through Friday.
The high usage alert is different from other alerts in that it uses a sophisticated algorithm to detect the user’s anomalous usage and triggers the email to alert the user of such an event, shortly after it occurs. This is not a recurring email on a set schedule.
This email is only triggered for consumption anomalies and so the user will not receive these emails if their usage patterns are generally consistent over the billing cycle.
A user can also go to their preferences page on the web portal directly or by clicking on a call-to-action button in email and unsubscribe from there.
Users will receive this email whenever they have a high consumption day that deviates from similar day usage profile. These similar day profiles will update over time, bill cycle over bill cycle, so if their consumption profile changes due to lifestyle changes, they will eventually stop receiving this email once their usage profile pattern is re-established by the algorithm.
A user can also go to their preferences page on the web portal directly or by clicking on a call-to-action button in email and unsubscribe from there.
The algorithm needs to establish a pattern differential to the candidate day with at least 5 similar days throughout the billing cycle of the user. Weekdays and weekends
are compared separately. On a given day, if the user’s energy consumption deviates significantly from the established similar day pattern, the high usage alert is triggered. The algorithm excludes vacation days or days with high temperature deviations.
Let’s call the day for which a high usage event is being calculated the candidate day. The candidate day’s energy consumption data is used as an input to the high usage algorithm which compares the candidate day consumption profile with a pattern established by at least 5 similar days within the billing cycle. Weekdays and weekends are compared separately. When the user’s candidate day consumption deviates from similar day usage patterns, a high usage alert is sent out to the user.
The High Usage Event Detection algorithm aims to identify days of unexpectedly high usage beyond the typical user behavior. We have developed a data driven approach to identify such days that works globally and personalizes the detection methodology to each user.