Appliance Analysis is the addon feature of the Ohm Assistant, with detection based on ML predictive models. Thus, sustained proper use is required for the ML-Bot learning to enhance accuracy. The Ohm Bot requires around 15 days of learning, and this is dependent on
- Bot Online: Quality data collection, using steady always-on Wi-Fi connection. Disruptions lead to data loss.
- Proper Appliance Activity Feedback: Feedback for all events, with care taken to ensure correct feedback.
- Note that the larger the house & greater number of appliances, the learning process will be more complex.
BOT CONFIDENCE & APPLIANCE ACTIVITY FEEDBACK
On the Analysis Main Screen widget, click on VIEW to access this section.
1. BOT CONFIDENCE (Right Hand Side Tab): This indicates the prediction accuracy in terms of Hugh, Medium & Low, across the top appliance categories. This will indicate the relative confidence of the bot analysis, based on the level of learning achieved, since the analysis works on the predictive model.
2. APPLIANCE ACTIVITY FEEDBACK (Left Hand Side Tab): This gives near real-time prediction of appliance ON events. Users should improve the bot confidence by providing feedback for each event. The ML-Bot requires sufficient training to enable it to differentiate between similar assets, hence it seeks validation of specific events to build up the specific machine learning (ML) model that applies to that home.
3. APPLIANCE ACTIVITY EVENTS: Here the Appliance Activity events are listed here. The category, date & time of commencement of operation of the appliance is displayed. The timestamp when the appliance is switched on is displayed here.
4. APPLIANCE ACTIVITY EVENT CATEGORIES: Appliance Activity events are displayed at category level for Air Conditioners, Geysers, Washing Machines, Heating Devices & Dishwashers. They are NOT displayed for Fridges, Electric Vehicles (EV), Pumps, Miscellaneous or Other Appliances.
5. EVENTS DISPLAYED AFTER TIME LAG: Each specific Appliance Activity detection event would be displayed up to 2 hours after the event. The aggregation on hourly basis is focussed on improving analytical accuracy, and if the interval of aggregation were to be reduced, the accuracy would be lower.
6. EVENTS SELECTIVELY DISPLAYED: Ohm displays only selected appliance activity events, generated automatically, for selected events for which the system seeks user validation. Hence each and every detection event would not be displayed. Thus, appliance activity is NOT designed to provide information on each event of use for each appliance. Events are displayed for only the last 48 hours, and can be filtered by category.
7. PROVIDING APPLIANCE ACTIVITY FEEDBACK FOR EACH EVENT:
On a day-to-day basis, users must pay special attention to each Appliance Activity Event that is displayed and provide required feedback for each activity that is displayed. This is essential to ensure proper training of the Ohm Bot through the Machine Learning process, to improve the quality of detection & Bot Confidence.
Therefore, the Appliance Activity list must be visited by the user regularly and cleared by providing the required feedback for each Appliance Activity event that is displayed. Click on the particular event under Appliance Activity and provide proper feedback.
8. PROPER DETECTION - TAG YOUR APPLIANCE: If the specific event is a proper detection, proceed to confirm the same by clicking TAG YOUR APPLIANCE. Tag the specific appliance for which the detection event is displayed.
With TAG YOUR APPLIANCE, the user can provide superior & sharper feedback and map the specific detection event to the specific appliance (and this complements the Train Your Bot functionality).
The selected option is highlighted in RED. Click on SAVE CHANGES to submit the details.
9. USE I’M NOT SURE TO AVOID GUESS WORK AND KEEP FEEDBACK ACCURATE: Appliance Activity feedback is an irreversible process; therefore, feedback must be accurate and given only when the user is confident about the event. Hence FALSE FEEDBACK must not be submitted, since this can severely affect the ML model being developed through the ML-Bot learning process. Hence at any point, if unsure about the event, click on I’M NOT SURE, rather than making a wrong guess.
The selected option is highlighted in RED. Click on SAVE CHANGES to submit the details.
10. FEEDBACK ACCEPTED: When the feedback is accepted & changes saved, the acknowledgment is displayed.
11. WRONG DETECTION: If the event does not seem to match expected appliance usage, click WRONG DETECTION.
(a) I’M NOT SURE: Select this if unsure about the appliance connected with the event, and ensure that FALSE FEEDBACK is not submitted, rather than making a wrong guess.
(b) SELECT THE APPLIANCE/CATEGORY: The list of other appliance categories is displayed. Navigate to the specific category. Navigate to the appliance list under that specific category, and select the specific appliance, only if certain.
The selected option is highlighted in RED. Click on SAVE CHANGES to submit the details.
12. OPERATING LIMITS OF OHM ASSISTANT PREDICTIVE ANALYSIS: The detection values that the bot indicates are predictions and not the actual consumption of the appliances. Ohm operates at an average detection accuracy of 85%. The accuracy may be affected depending on the size of the house & number of appliances. The larger the house and higher number of appliances affects the detection. Depending on the size of house and number of appliances, some appliances may not get classified correctly.
13. INCOMPLETE DATA
(a) No Data (Appliance Activity Events) in Appliance Activity (Bot is learning) - Displayed when feedback received for all listed events (of last 48 hours) & event list clear.
(b) No Data (Categories) in Bot Confidence (Page Blank) - Appliances are not added through Vyas.
NEIGHBOURHOOD:
On the Analysis Main Screen widget, click on VIEW to access this section.
The Neighbourhood comparison feature is designed to help understand your performance compared to your neighbours based on energy consumption. Tap on the circle (point) on the graph to see the daily/monthly rank and more details. The consumption is displayed in units, and the lower the value the better the consumption.
Statistics are drawn from the system for comparison (a) Average of all neighbours (b) The most efficient neighbour.
A score of ZERO indicates incomplete information updated in the Vyas and MyBots sections.
~Thank You~