Time Series, Dense Representations, and Sparse Representations of Temporal Data

A time series is an ordered sequence of measurements of a variable arranged according to time of occurrence. Time series are typically measured at a constant frequency with data points are spaced at uniform time intervals.

The characteristic properties of a time series include the following:

The data points are not independent.
The dispersion of data points varies as a function of time.
The data frequently indicates trends.
The data tends to be cyclic.

Typical business applications for time series analysis include the following:

Budgetary analysis
Economic forecasting
Inventory analysis
Process control
Quality control
Sales forecasting
Stock market analysis
Workload projections
Yield projections

The EXPAND ON clause enables forms of time series expansion on a PERIOD column value of an input row by producing a set of value-equivalent rows, one for each granule in the specified time period. The number of granules is defined by the anchor name you specify for the clause.

You can expand sparse PERIOD representations of relational data into a dense representation of the same data. Data converted to a dense form can be more easily manipulated by complex analyses such as moving average calculations without having to write complex SQL requests to respond to business questions made against sparse relational data.

The available forms of time series expansion for the EXPAND ON clause are the following.

Interval expansion, where rows are expanded by user-specified intervals.
Anchor point expansion, where rows are expanded by user-specified anchored points.
Anchor PERIOD expansion, where rows are expanded by user-specified anchored periods.

Time Series, Dense Representations, and Sparse Representations of Temporal Data - Teradata Vantage

Teradata® VantageCloud Lake