Natural language details for time series

Natural language details for time series in IBM® Cognos Analytics provides insights for time series data that is displayed in all applicable exploration visualizations.

When time fields and a measure field are specified in the visualization and a forecasting model is computed, detected time series insights become available under the Details tab.

Cognos Analytics supports three types of insights for time series data: unusual values, seasonal effects, and trend. Unusual values list observations that are statistically different from values that are predicted by the selected forecasting model. The seasonal effects insight provides a seasonal length, which is the duration of a seasonal pattern, for a time series. For example, average temperature variation across 12 months establishes an annual pattern. This insight also provides the strength of the seasonal effects and the time points with the largest and the smallest seasonal influence per seasonal pattern. The trend insight detects an overall direction of the time series together with its strength.

Insights for time series are based on the selected exponential smoothing model for time series data in the visualization. A forecasting model is computed regardless of the visualization type or activation of the forecasting feature in the visualization. Time series points are automatically sorted in chronological order for the insights detection. Unlike in the forecasting feature, the time points that are displayed in the visualization are not sorted.

Insights for time series expand on the existing insights under the Details tab. The latter treat a time field as a categorical field and provide basic summaries and insights that correspond to the visualization insights feature.

For more information, see Natural language details for time series.