Axibase Time Series Database
  • Overview

  • Series

  • Properties

  • Messages

  • Alerts

  • Metric

  • Entity

  • Entity Group

  • Replacement Table

  • Extended

  • API Clients

Interpolation

Overview

Interpolation transforms an input time series to a regularized series by calculating values at evenly spaced intervals using a linear or step function.

The interpolation process performed by the database is outlined below:

  1. Load samples for the selection interval specified with startDate and endDate parameters.
  2. If OUTER boundary mode is enabled, load one value before and one value after the selection interval to interpolate leading and trailing values.
  3. Create evenly spaced timestamps within the selection interval. Timestamps can be aligned to a calendar or start/end time of the selection interval.
  4. For each timestamp, calculate the value from the two nearest neighbor samples using linear or step interpolation function.
  5. If fill parameter is enabled, add missing leading and trailing values.

Fields

Name Type Description
period object [Required] Time interval used for interpolation.
function string [Required] PREVIOUS, LINEAR, or AUTO.
boundary string Enable values outside of the selection interval.
fill string Creates missing leading and trailing values.

period

Period is a repeating time interval used to create evenly spaced timestamps.

Examples:

  • { "count": 1, "unit": "HOUR" }
  • { "count": 15, "unit": "MINUTE", "align": "START_TIME" }
  • { "count": 1, "unit": "DAY", "align": "CALENDAR", "timezone": "US/Pacific" }

function

Name Description
LINEAR Calculate interpolated value by adding the difference between consecutive values proportional to their time difference.
PREVIOUS Set interpolated value equal to the previous value.
AUTO Apply the interpolation function specified by the metric interpolate setting. Default: LINEAR.

Notes:

  • Detailed values with timestamps equal to interpolated timestamps are included in the response unchanged.
  • The LINEAR function returns an interpolated value only if both the preceding and the following value is present.
  • The PREVIOUS function requires a preceding value. The last detailed value is used to calculate a final interpolated value in the response.

boundary

Name Description
INNER [Default] Data outside of the selection interval is not loaded by the database.
OUTER One value before and one value after the selection interval is loaded by the database to interpolate leading and trailing values.

Examples:

  • { "boundary": "OUTER" }

fill

The fill parameter eliminates gaps at the beginning and end of a selection interval. If boundary is OUTER and there are values on both sides of the selection interval, the fill parameter is not applied.

Name Description
false [Default] Do not add missing values.
true Add missing leading values by setting the values equal to the first available detailed value.
Add missing trailing values by setting the values equal to the last available detailed value.
{n} Add missing leading and trailing values by setting the values equal to the specified number {n}.
The number {n} can be any decimal number as well as NaN string (Not a Number).

Examples:

  • { "fill": false }
  • { "fill": true }
  • { "fill": 0 }
  • { "fill": "NaN" }

Examples

Dataset:

series e:nurswgvml007 m:cpu_busy=-1 d:2016-12-31T23:30:00Z
series e:nurswgvml007 m:cpu_busy=0  d:2017-01-01T00:30:00Z
series e:nurswgvml007 m:cpu_busy=2  d:2017-01-01T02:30:00Z
series e:nurswgvml007 m:cpu_busy=3  d:2017-01-01T03:30:00Z

| datetime         | value |
|------------------|-------|
| 2016-12-31 23:30 | -1    |
| 2017-01-01 00:30 | 0     |
| 2017-01-01 01:30 | ...   | -- Sample at 01:30 is missing.
| 2017-01-01 02:30 | 2     |
| 2017-01-01 03:30 | 3     |

Fill Gaps with LINEAR Function

[{
  "startDate": "2017-01-01T00:00:00Z",
  "endDate":   "2017-01-01T05:00:00Z",
  "entity": "nurswgvml007",
  "metric": "cpu_busy",
  "interpolate" : {
    "function": "LINEAR",
    "period": {"count": 1, "unit": "HOUR"}
  }
}]

Response:

With default INNER mode, values outside of the selection interval are ignored.

| datetime         | value |
|------------------|-------|
| 2017-01-01 01:00 | 0.5   |
| 2017-01-01 02:00 | 1.5   |
| 2017-01-01 03:00 | 2.5   |

[{"entity":"nurswgvml007","metric":"cpu_busy","tags":{},"type":"HISTORY","aggregate":{"type":"DETAIL"},
"data":[
    {"d":"2017-01-01T01:00:00Z","v":0.5},
    {"d":"2017-01-01T02:00:00Z","v":1.5},
    {"d":"2017-01-01T03:00:00Z","v":2.5}
]}]

LINEAR Function: 30 Minute Period

[{
  "startDate": "2017-01-01T00:00:00Z",
  "endDate":   "2017-01-01T05:00:00Z",
  "entity": "nurswgvml007",
  "metric": "cpu_busy",
  "interpolate" : {
        "function": "LINEAR",
        "period": {"count": 30, "unit": "MINUTE"}
    }
}]

Response:

| datetime         | value |
|------------------|-------|
| 2017-01-01 00:30 | 0.0   |
| 2017-01-01 01:00 | 0.5   |
| 2017-01-01 01:30 | 1.0   |
| 2017-01-01 02:00 | 1.5   |
| 2017-01-01 02:30 | 2.0   |
| 2017-01-01 03:00 | 2.5   |
| 2017-01-01 03:30 | 3.0   |

Fill Gaps with PREVIOUS Function

[{
  "startDate": "2017-01-01T00:00:00Z",
  "endDate":   "2017-01-01T05:00:00Z",
  "entity": "nurswgvml007",
  "metric": "cpu_busy",
    "interpolate" : {
        "function": "PREVIOUS",
        "period": {"count": 1, "unit": "HOUR"}
    }
}]

Response:

With default INNER mode, values outside of the selection interval are ignored.

| datetime         | value |
|------------------|-------|
| 2017-01-01 01:00 | 0.0   |
| 2017-01-01 02:00 | 0.0   |
| 2017-01-01 03:00 | 2.0   |
| 2017-01-01 04:00 | 3.0   |

LINEAR Interpolation with OUTER Boundary

[{
  "startDate": "2017-01-01T00:00:00Z",
  "endDate":   "2017-01-01T05:00:00Z",
  "entity": "nurswgvml007",
  "metric": "cpu_busy",
  "interpolate" : {
        "function": "LINEAR",
        "period": {"count": 1, "unit": "HOUR"},
        "boundary": "OUTER"
    }
}]

Response:

With OUTER mode, values outside of the selection interval are used to interpolate leading and trailing values.

| datetime         | value |
|------------------|-------|
| 2017-01-01 00:00 | -0.5  |
| 2017-01-01 01:00 | 0.5   |
| 2017-01-01 02:00 | 1.5   |
| 2017-01-01 03:00 | 2.5   |

LINEAR Interpolation with START_TIME Alignment

[{
  "startDate": "2017-01-01T00:15:00Z",
  "endDate":   "2017-01-01T05:00:00Z",
  "entity": "nurswgvml007",
  "metric": "cpu_busy",
  "interpolate" : {
        "function": "LINEAR",
        "period": {"count": 1, "unit": "HOUR", "align": "START_TIME"}
    }
}]

Response:

| datetime         | value |
|------------------|-------|
| 2017-01-01 01:15 | 0.75  |
| 2017-01-01 02:15 | 1.75  |
| 2017-01-01 03:15 | 2.75  |

LINEAR Interpolation, Leading/Trailing Values Filled

[{
  "startDate": "2017-01-01T00:00:00Z",
  "endDate":   "2017-01-01T05:00:00Z",
  "entity": "nurswgvml007",
  "metric": "cpu_busy",
  "interpolate" : {
        "function": "LINEAR",
        "period": {"count": 1, "unit": "HOUR"},
        "fill": true
    }
}]

Response:

| datetime         | value |
|------------------|-------|
| 2017-01-01 00:00 | 0.0   |
| 2017-01-01 01:00 | 0.5   |
| 2017-01-01 02:00 | 1.5   |
| 2017-01-01 03:00 | 2.5   |
| 2017-01-01 04:00 | 3.0   |

LINEAR Interpolation, Leading/Trailing Values Filled with NaN

[{
  "startDate": "2017-01-01T00:00:00Z",
  "endDate":   "2017-01-01T05:00:00Z",
  "entity": "nurswgvml007",
  "metric": "cpu_busy",
  "interpolate" : {
        "function": "LINEAR",
        "period": {"count": 1, "unit": "HOUR"},
        "fill": "NaN"
    }
}]

Response:

| datetime         | value |
|------------------|-------|
| 2017-01-01 00:00 | null  |
| 2017-01-01 01:00 | 0.5   |
| 2017-01-01 02:00 | 1.5   |
| 2017-01-01 03:00 | 2.5   |
| 2017-01-01 04:00 | null  |