Forecasting Time Series by Selecting the Most Suitable Model

If you need to forecast sales, budget or any other related figures, feel free using this widget, which will execute more than 15 statistical models and forecast results based on the best possible model.

The selected model is the one having lower error metrics among most of the executed models.

Why Time Series

  • Anything that is observed sequentially over time is a time series.
  • Time series forecasting models use only information based on one variable: the one to be forecasted, and makes no attempt to discover the factors which affect its behavior. It will extrapolate trend and seasonal patterns, but will ignore all other information such as marketing initiatives, competitor activity, changes in economic conditions, and so on.
  • Regular Time Series: The models executed by the widget have as input regular time series, e.g. time series that are observed at regular intervals of time (quarterly or monthly).

Forecasting Time Series

Time series models implemented by this utility execute a wide range of models, from 'naive' ones thru exponential smoothing and ARIMA models.

How to use this widget
  1. Choose the starting date of the time series.
  2. Choose how many periods to forecast (bear in mind that the longer the horizon, the wider the uncertainity of the forecast)
  3. Enter the time series input, separated by commas.
    For example:
    2787, 3891, 3179, 2011, 1636, 1580 ,1489, 1300 ,1356, 1653
  4. Specify if this is a Monthy or Quarterly time series.
    In the previous example, the time series data is Quarterly

Be patient since it may take a few seconds till you get the results...