Hi Support,
I would like to clarify how trend data is calculated and displayed.
From my understanding, aggregation means that multiple 5‑minute average data points are grouped into a single 1‑hour data point by calculating the hourly average, minimum, and maximum values.
Kindly advise on the following:
1. For historical data (e.g. whole March 2026), does the trend graph use 1‑hour data or 5‑minute data?
2. If it uses 1‑hour data, does that mean each hour contains only 1 data point instead of 12 data points (5‑minute intervals) and all the minutes seconds in the graph will show the same value of datapoints?
3. When zooming into the graph, does the system still display 1‑hour data points, or does it display to 5‑minute data?
4. How does the system handle missing data within an interval (e.g. timeout or no readings)? Is it ignored, shown as null, or interpolated?
Thank you for your clarification.
Hi Asri,
1. The average trending should show 5 minute averages.
2. Yes, it means that only 1 datapoint is stored instead of 12 and the graph should show the same value of datapoints. if the average trending is storing 1 year worth of data it would be 1hour averages.
3. It should stick to 1hour data points.
4. Exception values are not part of the consideration of the average trending and will be ignored.
You can find the interval used for the averaging with respect to the duration of the trend recorded as well as more info on trending here: https://docs.dataminer.services/dataminer/Operator_guide/Trending/Trending_techniques.html
If showing missing data/looking for abnormalies within the trend graph, I would suggest using real-time data to capture these abnormalies. Do take note of its shorter Time to Live.
It is also possible to record both average and realtime, and that will allow you to check for abnormalies using real-time trending and keep a longer histroric record of your trending.
I hope this helps.
Hi,
That is correct.
1. You indeed need to change this minute records to 25 weeks (or maybe take 26 weeks which is half of one year = 52 weeks)
2. Hour records are completely separate and independent of each other. You can leave this one as it was, or you can change it, it doesn't matter.
Note: when looking at trend data in Cube, the UI will automatically do averaging in the client app as well. This means that when zooming out, Cube will automatically start to calculate and show averaged data, and that could even be 15 minute time slots or any other number. To disable this, you can right click on the trend graph, and activate the option "Show most detailed data". Then it shows the most detailed time slot from the database directly without interfering with the data. It is possible that the graph will react slower under certain circumstances when activating this option.
Bert
Hi Frans,
Thank you for the update.
I would like to seek further clarification regarding the trending settings.
Currently, our system is configured as follows:
* Real-time: 1 day
* Minute records: 4 weeks 3 days
* Hour records: 52 weeks 2 days
We have received a requirement from the client to visualize 5-minute trend data for up to 6 months.
From our understanding, this would require extending the Minute records period accordingly.
At present, the 5-minute trend data is only available for approximately one month based on the current configuration.
We would like to confirm the following:
1. If we would like to retain and view the minute records for up to 6 months, we should extend the retention from 4 weeks to approximately 25 weeks, correct?
2. If the Minute records retention period is extended, would it also be necessary or recommended to adjust the Hour records retention settings accordingly?
We would appreciate your advice on the recommended best practice for this scenario.
Thank you.