Fit Time Series

The Fit command of the Time Series menu will fit a time series model to an existing set of data; much like the process for fitting distributions, the time series fitting process will attempt to match a selected set of time series models to an existing time series data set. However, unlike the distribution fitting process, the time series fitting process takes into account the internal structure, such as autocorrelation, trending, or seasonal variations), of the data points used in the fit.

There are a few requirements for fitting time series data:

  1. The data set must include at least six data points.
  2. The cells selected for use in the fitting process must be contiguous.
  3. All sample values must fall in the range -1E+37 <= X <= +1E+37.

Once a time series model has been fit it can be added to an @RISK model as a set of inputs (depending on the number of elements created during the fitting process) for use during simulations.

Figure 1 - Time Series Fitting

Time Series Fitting Window

The Time Series Fitting window has two tabs:

  • Data - Define the time series data to be included in the fitting process, as well as any transformations and other fitting options to be used. See Fit Time Series Data for more information.
  • Models to Fit - Select the time series distributions that should be tested during the fitting process.