Are there any tools directly abailable in InfluxDB to make some kind of predictive analysis with incoming queries?
(e.g. ARIMA Model)
Hello @Manfred,
There ARIMA isn’t supported natively, but holt winters is:
Which can be converted to double exponential smoothing as well.
Hello,
the referenced example for holtwinters is working now:
In the first step data is imported with this flux command:
import “experimental/csv”
csv.from(url: “https://influx-testdata.s3.amazonaws.com/noaa.csv”)
2nd the holt winter function is applied e.g.:
from(bucket: “NOAA”)
|> range(start: v.timeRangeStart, stop: v.timeRangeStop)
|> filter(fn: (r) => r[“_measurement”] == “h2o_feet”)
|> filter(fn: (r) => r[“_field”] == “water_level”)
|> filter(fn: (r) => r[“location”] == “santa_monica”)
|> aggregateWindow(every: 30m, fn: mean, createEmpty: false)
|>holtWinters(
n: 100,
seasonality: 4,
interval: 379m,
withFit: false,
timeColumn: “_time”,
column: “_value”,
)
The holtwinters prediction is shown in the dashboard.
But the prediction values don’t really fit to the real values.
I modified the paramteres in the holdwinters()-function and aggregateWindow.
Anyhow the result was not satisfiying.
Any suggestion how to opmimize that?
