Could you please let me know if there is an alternative to New Relic's EXTRAPOLATE command? This command is used to fill in missing gaps when calculating time-series data etc, ensuring smooth and continuous data representation.
To clarify the use case of EXTRAPOLATE and avoid confusion, I’ve included examples below.
- What does it do?
- It estimates (extrapolates) missing data points to fill in gaps, ensuring your calculations (like averages or rates) are as accurate as possible.
- This is especially useful if your app or data collection intermittently stops reporting events (e.g., network issues).
- Why use it?
- Without extrapolation, gaps in data can skew your results, making them artificially low.
- With extrapolation, New Relic assumes the trend continues during the gap, providing a more accurate estimate.
Example Without EXTRAPOLATE:
Suppose you collect data for 1 hour, but your app fails to send data for 15 minutes. Your dataset might look like this:
Time Bucket. Count (count(*))
12:00 - 12:15 PM 10
12:15 - 12:30 PM 0
12:30 - 12:45 PM 0
12:45 - 1:00 PM. 10
Without extrapolation:
- Total Count = 10+0+0+10=20
- Rate = 20/1 hour=20 events/hour (lower than actual).
Example With EXTRAPOLATE:
New Relic uses the available data to fill in gaps by assuming consistent reporting during the missing time:
Time Bucket. Count (count(*)) Extrapolated Count
12:00 - 12:15 PM 10 10
12:15 - 12:30 PM Missing 10 (estimated)
12:30 - 12:45 PM Missing 10 (estimated)
12:45 - 1:00 PM 10 10
With extrapolation:
- Total Count = 10+10+10+10=40
- Rate = 40/1 hour=40 events/hour (closer to the actual rate).
