Forecasting using SAS Forecast Server, SAS/ETS, and more

Question on ARIMA Time Series

Reply
Frequent Contributor
Posts: 118

Question on ARIMA Time Series

Hi,

 

I have a question on ARIMA Time Series model. 

 

It is Hospitality industry historical booking data. If a customer make a booking, then Type = 'Booking'; If a customer cancelled the booking, then Type = 'Cancel'. And its original 'Booking' disappears. 

 

Demand forecasting model uses ARIMA Time series model, and it uses historical demand as input only where Type = 'Booking' (Not using cancellation) Does it mean the output for demand forecatsing only inlude pure 'Booking' not 'Cancelation'? 

 

Thank you! 

Trusted Advisor
Posts: 1,228

Re: Question on ARIMA Time Series

Hi,

 

This question is related to logic how to define variable being used for ARIMA model. If you are taking only "booking" then consider those customers who booked and showed up for further analysis.

Frequent Contributor
Posts: 118

Re: Question on ARIMA Time Series

Thank you! Which means if I only take 'Booking' then forecasting output should only be 'Booked and Showed up' for future days?
Trusted Advisor
Posts: 1,228

Re: Question on ARIMA Time Series

I mean this would be based on your requirement. If a customer booked and then cancelled resulting no transaction for analysis.

Frequent Contributor
Posts: 118

Re: Question on ARIMA Time Series

Thank you for your reply! Sorry I didn't explain very clearly. 

 

In my situation, if a customer booked and then canncelled, the row of 'Booking' disappears and a new row of 'Cancel' comes out. 

 

If I just used 'Booking' as input, shall I assume my output for demand forecasting will only include 'Booking' part that already excludes cancellations if the model is good. 

Trusted Advisor
Posts: 1,228

Re: Question on ARIMA Time Series

What is the cancellation rate? Also, what is time interval for the time sereies and why do you think ARIMA is a good model for your data?

Ask a Question
Discussion stats
  • 5 replies
  • 207 views
  • 0 likes
  • 2 in conversation