Data Downsampling (LTTB)
Over a long time range, the number of records can be far larger than the number of pixels a chart can show on screen. In that case the chart is drawn with fewer points by using the Aggregation setting on the Reporting screen. When Largest-Triangle-Three-Bucket (LTTB) is selected as the Aggregation Method, the number of points is reduced while the visual shape of the data is kept.
LTTB affects only how the chart is drawn. Records in the archive, the data table and exported data do not change.
How It Works
The first and last points are kept. The first and last records of the series always appear in the chart.
The data is split into buckets. The records in between are divided into groups (buckets) of roughly equal size, based on the target number of points.
One point is selected from each bucket. For every point in a bucket, the area of the triangle formed by these three points is calculated:
- the point selected from the previous bucket,
- the point being examined,
- the average of the points in the next bucket.
The point that forms the largest triangle represents that bucket.
The process is repeated from left to right. The selection in each bucket depends on the point selected in the previous bucket.
The points that form the largest triangles are the peaks, dips and sudden changes of the curve. This is why LTTB keeps short spikes visible in the chart, which methods such as Average can smooth out.
Which Method When
| Goal | Recommended Aggregation Method |
|---|---|
| See the shape of the curve, its peaks and dips | Largest-Triangle-Three-Bucket |
| See the average for a period | Average |
| See the worst case (lower or upper limit) | Minimum or Maximum |
| See the total of a counter or quantity | Sum |
When exact values are needed, turn Aggregation off or use the data table.