TD_SIGNIF_PERIODICITIES Function | Teradata Vantage - TD_SIGNIF_PERIODICITIES - Teradata Vantage

Database Unbounded Array Framework Time Series Functions

Deployment
VantageCloud
VantageCore
Edition
Enterprise
IntelliFlex
VMware
Product
Teradata Vantage
Release Number
17.20
Published
June 2022
Language
English (United States)
Last Update
2024-10-04
dita:mapPath
ncd1634149624743.ditamap
dita:ditavalPath
ruu1634160136230.ditaval
dita:id
ncd1634149624743

TD_SIGNIF_PERIODICITIES is a statistical test to determine if there exist any significant periodicities (seasonal cycles) in the residual series. This test can be used with any residual series, though it is most often used with the residuals produced by TD_ARIMAESTIMATE with FIT_PERCENTAGE(100) or TD_ARIMAVALIDATE.

Significant periodicities are recurring patterns or trends that occur at regular intervals in time series, financial data, and sensor data. Detecting periodicities is important for understanding the underlying trends and cycles that govern the data. By identifying them, data scientists can develop models and algorithms that can capture and analyze these patterns to improve accuracy in forecasting the data. It can also help identify anomalies or unusual events in the data, and lead to corrective actions.

Detecting significant periodicities in time series data is crucial for accurate forecasting, especially in industries such as retail or finance, where it impacts decision making. By accounting for seasonal effects, businesses can adjust their inventory or marketing strategies to meet demand better. Additionally, identifying significant periodicities in the residual series can provide insights into the underlying patterns of the data.

Detecting significant periodicities in the residual series is essential for modeling and understanding time series data. It improves forecasting accuracy, provides insights into underlying data patterns, and aids decision making. Accounting for seasonal effects and validating models, data scientists can make informed decisions and lead to better business outcomes.

The following procedure is an example of how to use TD_SIGNIF_PERIODICITIES:
  1. Use TD_ARIMAESTIMATE with FIT_PERCENTAGE less than 100 to generate an ART table containing an ARIMA estimate model.
  2. Use TD_ARIMAVALIDATE to validate the model produced by TD_ARIMAESTIMATE..
  3. Use TD_EXTRACT_RESULTS to retrieve residuals from the TD_ARIMAVALIDATE result.
  4. Use TD_LINESPEC or TD_POWERSPEC with FREQ_STYLE parameter set to K_PERIODICITY to perform spectral analysis on the ARIMA validate residuals.
  5. Use one of the following methods to identify the top four periodicities returned from the line spectrum or power spectrum analysis:
    • Issue a query similar to the following to determine top four periodicities.
      SELECT TOP 4 * FROM LINESPEC_VALIDATE_AR2_RESIDUAL ORDER BY SPECTRAL_DENSITY_RESIDUAL DESC;
    • Use TD_PLOT to plot results of line spectrum and power spectrum. Visually determine the top four periodicities by identifying the top four peaks.
  6. Use TD_SIGNIF_PERIODICITIES on the periodicities of interest. More than one periodicities can be entered using the PERIODICITIES parameter.
  7. Use TD_SIGNIF_PERIODICITIES passing in the list of periodicities identified in the previous step.
    More than one periodicity can be entered using the PERIODICITIES parameter for the periodicities of interest.