Forecasting/Predictions – Documentation and Support | Knowi

Cloud9QL can forecast and predict future data points for any dataset, based on its historical values.

To predict a single point in the future:

select PREDICT(<field to predict>, <date field>, <start date for prediction>[, <prediction model>])

To predict multiple data points in the future:

select PREDICT(<Field to Predict>, <Date Field>, <Start Date for Prediction>,
    <Prediction Frequency>, <Prediction Data Points>[, <prediction model>])

To predict multiple data points in the future based on secondary dimension(s) (ie: grouping(s)):

select PREDICT(<Field to Predict>, <Date Field>, <Start Date for Prediction>,
    <Prediction Frequency>, <Prediction Data Points>[, <prediction model>])
group by <Dimension 1>, ..., <Dimension N>

PREDICT loads the historical data points, determines the optimum prediction model, and returns the predicted future data point(s).

Parameters:

<prediction frequency>: the spacing between predicted points, in the format <Number><TimeUnit>. For example, 1d means daily and 2m means every 2 months.
<prediction model>: optional. By default the optimum model is selected automatically. To force a specific model, use one of the following (case-sensitive): Regression, PolynomialRegression, MultipleLinearRegression.

Example:

select predict(Count, Date, 07/10/2015, 1d, 5)

Using input data:

Date, Count
07/01/2015, 1
07/01/2015, 2
07/02/2015, 4
07/04/2015, 8
07/07/2015, 14

Related articles