TD_LINESPEC is one of a set of functions to perform spectral analysis of a series. It is used to identify cycles that may be inherent in an input series. The identified cycles are expressed in terms of an integral frequency, radians, a sample rate, a frequency (HERTZ), or a periodicity.
Spectral analysis of a time series provides insights into the underlying patterns or cycles in the data. It is used to identify periodicities or oscillations in the data, such as daily, weekly, or seasonal cycles. It can also detect trends, abrupt changes, or other features of interest in the frequency domain.
The spectral analysis of a time series involves analysing the frequency content of a time series in the frequency domain. It parses a time series into its component frequencies, and measures the amplitude or strength of each frequency component. The resulting frequency spectrum can be visualized using a power spectral density plot or a periodogram.
Spectral analysis is used in physics, engineering, economics, and finance, where time series data is common. It helps in understanding the behavior of complex systems, forecasting future trends, and developing mathematical models for prediction and control.
- Use TD_ARIMAESTIMATE to generate an ARTFITRESIDUALS layer.
- Use TD_EXTRACT_RESULTS to retrieve residuals from the TD_ARIMAESTIMATE ARTFITRESIDUALS layer.
- Use TD_LINESPEC with FREQ_STYLE parameter set to K_PERIODICITY to perform spectral analysis on the extracted residuals, and place the results in an ART named LineSpecResults.
- Use the following command to retrieve top five spectral density magnitudes:
SELECT TOP 5 * FROM LineSpecResults ORDER BY SPECTRAL_DENSITY_field DESC; - Use the retrieved top five list to form a PERIODICITIES parameter list, and then use TD_SIGNIF_PERIODICITIES to run a significant periodicities test against the list.
- Retrieve results of the TD_SIGNIF_PERIODICITIES test.