timeseries

Description

A library for handling timeseries with a fixed interval.

A timeseries consists of two parts. The data and meta information. Meta information includes

• Timeseries name: An arbitrary name for reference. It is a best practice to use the interval as suffix for the name (e.g. @costs_pt15m@)
• Interval
• Timezone name (e.g. "Europe/Berlin")

Grouping and Aggregation

Groups the timeseries by day and sums the value:

$interval = new DateInterval('P1M');
$chunkOperation = new \Flagbit\Timeseries\Operation\ChunkOperation($interval);
$chunks = $chunkOperation->split($timeseries);
$sumAggregation = new \Flagbit\Timeseries\Aggregation\SumAggregation();
$aggregatedValues = [];
foreach ($chunks as $chunk) {
    $aggregatedValues[] = $sumAggregation->aggregate($chunk);
}
$timeseries = new \Flagbit\Timeseries\Timeseries($timeseries->getStart(), $interval, $aggregatedValues);

Avaliable Aggregations are

• MAX
• MIN
• SUM

Grouping is currently only possible by DateTimeInterval. The Chunk Interval has to be a multiple of the timeseries' interval.

Querying and Persistence

The Persistence Layer is still under heavy development. So please refer to the "Future Plans" section for the conceptual thoughts.

Future Plans

• It MUST be possible to save a timeseries with an arbitrary name
• The library MUST select the best way to store the data
• The library MUST store timeseries with different intervals in different tables.
• The library MUST provide a way to select and update a timeseries only partially.
• The library MUST provide a way to split a timeseries at any given point in time.
• The library SHOULD "snap" to the "next best" interval if the given range on selecting a timeframe is not aligning with the timeseries' intervals.

See RFC2119.