Trend Resampling

Trend resampling is the process of transforming time-series data from one interval to another by aggregating values into larger time buckets. This is commonly used to reduce data resolution for visualization, reporting, or storage optimization.

Intervals

Resampling transforms data from one fixed interval to another. The original data must be uniformly sampled (e.g., every 15 minutes), and resampling groups this data into larger time buckets. If the selected interval matches the original, the data is returned unchanged.

Supported resample intervals:

• auto (automatically selects the best interval based on time span)
• 15min
• 30min
• 1hr
• 1day
• 1mo

Aggregation Functions

When resampling, multiple values from a smaller interval are combined into a single value for a larger interval. This is done using aggregation functions, which determine how the values within each group are reduced. Different types of data require different strategies:

• Use mean for smoothing or averaging sensor noise.
• Use sum for totals over time, like energy or flow.
• Use min/max to track range extremes.
• Use diff to measure change across the interval.

For most aggregation functions, the result is placed at the starting timestamp of the interval as seen in examples below. For diff the result is placed at the ending timestamp, since it represents the net change across the interval.

mean

The arithmetic average of all non-null values in the group.

Use case: Normalize noisy signals, reduce data size while preserving trends.

Example:

Timestamp	Values	mean
00:00	2.0	5.0
00:15	4.0
00:30	6.0
00:45	8.0

sum

The sum of all non-null values in the group.

Use case: Calculate total energy use, flow, or any cumulative metric over time.

Example:

Timestamp	Values	sum
00:00	2.0	20.0
00:15	4.0
00:30	6.0
00:45	8.0

min

The smallest non-null value in the group.

Use case: Track minimum temperatures, loads, or pressure within each interval.

Example:

Timestamp	Values	min
00:00	6.0	1.5
00:15	3.0
00:30	1.5
00:45	8.0

max

The largest non-null value in the group.

Use case: Identify peak values, such as maximum demand or temperature.

Example:

Timestamp	Values	max
00:00	6.0	8.0
00:15	3.0
00:30	1.5
00:45	8.0

diff

Calculates the difference between the last and first non-null values in the group.

Use case: Compute net change over time, such as delta kWh.

Example:

Timestamp	Values	min
00:00	2.0	null
00:15	4.0
00:30	6.5
00:45	8.0
01:00	10.0	8.0

NA Handling

Real-world data often includes missing values (na, meaning “not available”). During resampling, these gaps are handled consistently to ensure meaningful results while avoiding misleading calculations. The rules below define how na values are treated for each aggregation group.

• If all values in a group are na, the result is na.
• If some values are na, they are ignored in aggregation.
• For diff, the first and last non-null values are used. If there are not at least two non-null values, the result is na.