Make Equally Spaced
The Make equally spaced method converts one or more given time series to a uniform temporal scale. The generated equally spaced time series have the same start time as the initial time series.
Invoke the method via the menu entries Methods → Change time axis → Make equally spaced → Preserve integral/Interpolating. In contrast to the other method Make equally spaced, the menu item is used to select whether the method is to be integral-maintaining or interpolating (see following figure).
Preserve Integral
The integral-preserving normalization converts, for example, electricity demand measurements in 15-minute increments and gas meter reading records in 4-hour increments into a common hourly increment; or non-equidistant measurement intervals can be converted to equidistant intervals without changing the integral.
The integral-maintaining method assumes that the measured values are preliminary averages of the following period. This means that a point describes the integral of the interval following it. The following figure schematically illustrates the process of downsampling.
in the lower toolbar. First Select time series to be normalized (see following figure).
Then enter the desired step size of the target time series (Sample time new). Specify whether the existing time series is to be overwritten by the equally spaced time series or whether a new time series is to be created and, if necessary, provided with a Suffix.
Interpolating
An interpolation method is particularly suitable for temperatures and pressures.
The following figure schematically illustrates the process of interpolation in which data points are added to the time series.
First select time series to be interpolated and enter the time step size (Sample time new) in order to generate an equidistant time series by interpolating (see following figure). Specify whether the existing time series is to be overwritten by the equally spaced time series or whether a new time series is to be created and, if necessary, provided with a Suffix.
Select an Interpolation Method from the drop-down list (see figure below).
The interpolation methods Linear Interpolation and Cubic Spline Interpolation are available. The interpolation method has an effect only if the resulting time series has a finer grid and therefore more data points than the original time series. The following figure schematically illustrates the difference between the actual progression and the results of linear and spline interpolation.
Linear interpolation preserves all local and global extreme values of the original time series. If the actual extremes have not been recorded due to discretization, cubic spline interpolation can provide a more accurate approximation of the actual course.
Linear Interpolation
The values are determined by linear interpolation of the adjacent data points.
The following figure compares the generated equally spaced (top) with the original unequally spaced time series (bottom).
Cubic Spline Interpolation
Cubic spline interpolation calculates piecewise a third degree polynomial (cubic spline). The values are determined by calculating the cubic splines for each data point.
