Forecasting tickets


1 Comentarios

  • Comentario oficial
    Bryan Flaherty
    Zendesk team member

    Hey Evan!

    Before we setup the forecasting we should first double check that the data you are forecasting is the kind of data Explore is going to be good at forecasting.  Full forecasting models consider a wide variety of variables that would need to be tailored to the use case.  As the Explore forecasting model is designed to work mostly out of the box it only considers past time based trends.  Explore forecasting is going to be at its most accurate when you are looking at data that has a trend and then cyclical variation on that trend.

    If that sounds like your ticket data we can do the following steps:

    1. In Query Builder, create a time-based query. For example, you could add COUNT(Tickets) to the Metrics panel and Ticket created - Year and Ticket created - Week to the Columns panel.
    2. Click the result manipulation icon (the up and down arrow), then click Forecast.
    3. In the Forecast menu, check Calculate a forecast on the result
    4. For Path select column (as this is where we put the time based attribute)
    5. Under Method, choose either Additive trend and multiplicative season or Additive trend and additive season. (These are different ways of calculating the forecast, it will likely be worth experimenting to see what method better fits your data)
    6. Under Values to predict, choose 1
    7. Under Values per cycle, select Auto
    8. When you are finished, click Apply

    I hope this helps!  For more information on forecasting please see Using Forecasting to Predict Results

Iniciar sesión para dejar un comentario.

Tecnología de Zendesk