Predictive Analytics can be applied to any of your datasets to derive best-fit forecasts based on historical data.

This tests a given dataset against a variety of forecasting models to determine the best model best suited to the data based on historical data with the least Sum of Absolute Errors (SAE).

Models used behind the scenes include:

- Simple Exponential, Double Exponential, Triple Exponential Smoothing Models
- Moving Averages and Weighted Moving Averages
- Naive Forecasting Model
- Regression and Polynomial Regression Model
- Multiple Linear Regression Model

Predictions can be applied to any of your datasets (or during query time) using Cloud9QL.

To predict 1 point 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 point in the future:

```
select PREDICT(<Field to Predict>, <Date Field>, <Start Date for Prediction>,
<Prediction Frequency>, <Prediction Data Points>)
```

This:

- Loads the data points from input data.
- Determines the optimum prediction model.
- Predicts future data point(s).
- <prediction frequency> is in the format of <Number><TimeUnits> for example 1d means daily 2m specifies every 2 months.
- <prediction model> you can choose a specific model to be applied from one of the following supported models (case-sensitive):
- Regression
- PolynomialRegression
- MultipleLinearRegression
- MovingAverage
- SimpleExponentialSmoothing
- DoubleExponentialSmoothing
- TripleExponentialSmoothing

For advanced predictive modeling and machine learning algorithms, including using your own custom prediction models, see our machine learning capabilities.