Overview of the forecasting preview

The preview of the forecast consists of three main components.

The three main components are:
  1. Prediction accuracy indication.
  2. Preview chart. In a multivariate forecast, the preview chart includes tabs for each variable that is included in the forecast.
  3. Statistical details, which vary for univariate or multivariate forecasts.
Three main preview components

The illustrations on this page show a univariate forecast, but the concepts are the same for a multivariate forecast.

Prediction accuracy

Prediction accuracy

Prediction accuracy is one of the most important indicators of the usefulness of a given prediction and represents how well the model was able to fit the historical data. If the accuracy is indicated as low, then the exponential smoothing model that is identified was not a good fit for this data. The prediction is likely not usable and other means should be employed to derive this data.

Preview chart

The preview chart consists of the following three main areas:

  1. The confidence interval (1). This is not included in multivariate forecasts.
  2. All historical information that was used to create the model.
  3. And the forecasted data points based on the model that best aligned with the historical data.
Figure 1. Univariate forecast preview
Prediction accuracy

A multivariate preview also includes tabs for each of the variables that are used to calculate the forecast. Click a tab to view details.

Figure 2. Multivariate forecast preview
Prediction accuracy

Statistical Details - univariate forecasts

The statistical details for a univariate forecast includes information that is typically useful to a data scientist and consists of the following five main areas:

  1. The accuracy details, which include the number that is used to derive the overall accuracy of the forecast.
  2. Parameters.
  3. Information about the forecasting method that was used along with the detected trend and seasonality components.
  4. Trend and seasonality strength and seasonality period. Period indicates information that relates to the cycles detected in the data along the time series.

If either the trend component or the seasonal component is None, then the respective characteristic was not detected and does not provide a Strength value (or Period for seasonality).

Prediction accuracy

See Univariate statistical details for more information.

Statistical details - multivariate forecasts

The statistical details for a multivariate forecast consists of the four main areas.

Accuracy
Indicates the accuracy of the forecast based on available historical data.
Measures of error and model fit
Provides details of the factors that contribute to the accuracy prediction. For more information, see Multivariate statistical details.
Statistical model
Indicates which model was used to generate the forecast. See Multivariate forecasts for more information about how a model is chosen for a forecast.
Relative contribution
Lists the variables that are used to generate the forecast and indicates the relative contribution of each variable to th foirecast.
Statistical details page for a multivariate forecast