Forecasting Ridership Data

For technical questions regarding estimation of single equations, systems, VARs, Factor analysis and State Space Models in EViews. General econometric questions and advice should go in the Econometric Discussions forum.

Moderators: EViews Gareth, EViews Moderator

usi
Posts: 4
Joined: Fri Apr 16, 2021 7:25 am

Forecasting Ridership Data

Postby usi » Fri Apr 23, 2021 6:14 am

Hi, I am a civil engineer and beginner in eviews. I am trying to estimate future passenger flow data based on the past statistics. I do automatic ARIMA forecasting in eviews. However, the selected model gives a result in the figure attached. It is very different from the actual data. What am I doing wrong?

ARIMA Graph.png
ARIMA Graph.png (67.1 KiB) Viewed 4507 times

startz
Non-normality and collinearity are NOT problems!
Posts: 3775
Joined: Wed Sep 17, 2008 2:25 pm

Re: Forecasting Ridership Data

Postby startz » Fri Apr 23, 2021 7:33 am

Statistics is never as good as knowing about the subject. Using an ARIMA model just forecasts based on previous values. You need to bring your civil engineering knowledge to bear.

What do you think is wrong with the forecast?

usi
Posts: 4
Joined: Fri Apr 16, 2021 7:25 am

Re: Forecasting Ridership Data

Postby usi » Sat Apr 24, 2021 3:40 am

startz wrote:Statistics is never as good as knowing about the subject. Using an ARIMA model just forecasts based on previous values. You need to bring your civil engineering knowledge to bear.

What do you think is wrong with the forecast?


Hi,

Thanks for your response.

I am thinking that the forecast should not be steady without any variations in the data. The forecast I made shows a regular profile just with M shapes but same at any stage of the year, no variations during the whole year. Maybe I should do seasonal ARIMA (SARIMA) but I don't know if eviews can do that.

startz
Non-normality and collinearity are NOT problems!
Posts: 3775
Joined: Wed Sep 17, 2008 2:25 pm

Re: Forecasting Ridership Data

Postby startz » Sat Apr 24, 2021 6:59 am

EViews can do seasonal arima, although I think you have to give a specification rather than use the automatic feature. You can also use dummy variables for seasons.

While I don't know much about traffic, from your graph I have a suspicion that the idea "weekend" and "holiday" might be important.

usi
Posts: 4
Joined: Fri Apr 16, 2021 7:25 am

Re: Forecasting Ridership Data

Postby usi » Sat Apr 24, 2021 7:14 am

startz wrote:EViews can do seasonal arima, although I think you have to give a specification rather than use the automatic feature. You can also use dummy variables for seasons.

While I don't know much about traffic, from your graph I have a suspicion that the idea "weekend" and "holiday" might be important.


How can I give this kind of specification?

startz
Non-normality and collinearity are NOT problems!
Posts: 3775
Joined: Wed Sep 17, 2008 2:25 pm

Re: Forecasting Ridership Data

Postby startz » Sat Apr 24, 2021 7:41 am

@weekday gives the day of the week. You probably want to use the ls command to specify an ARIMA or other model including dummy variables for the day of the week or for the season of the year.

usi
Posts: 4
Joined: Fri Apr 16, 2021 7:25 am

Re: Forecasting Ridership Data

Postby usi » Sat Apr 24, 2021 8:03 am

How is SARIMA done?

startz
Non-normality and collinearity are NOT problems!
Posts: 3775
Joined: Wed Sep 17, 2008 2:25 pm

Re: Forecasting Ridership Data

Postby startz » Sat Apr 24, 2021 9:15 am



Return to “Estimation”

Who is online

Users browsing this forum: Cscottmcwall-ea and 26 guests