Level: Advanced
Job Function: Analyst / Scientist / Engineer
Industry: Cross Industry
Jian Cao, JMP Principal Systems Engineer, SAS
How can you model and forecast the response series when it is interrupted due to interventions or shocks (e.g., marketing campaigns, pricing changes)? If you have leading indicators or other exogenous series, how can you bring them as input variables into your normal ARIMA model to produce better forecasts? In this presentation I will demystify the transfer function models in the JMP Time Series platform through two real-life case studies: one for intervention analysis and the other for ARIMA with input variables. I will explain and illustrate the process of exploring the patterns in both output and input series, identifying the dynamic input-output relationship, and fitting and comparing transfer function models with different specifications.
