Theil Statistics Programme Command

[Saakshi]

Hello everyone,

Can someone really help me out? I badly need help!

I am doing out-of-sample forecast. Following is my program:

gg_outsmpl.prg

Out-of Sample Forecast for H horizon

(3.16 KiB) Downloaded 537 times

As a forecast evaluation test, I also want to check the Theil statistics. I have tried with @theil command but is unable to find the appropriate result.

Please help me out with the problem.

Best,
Saakshi

[EViews Gareth]

All of the forecast evaluation type statistics (@rmse, @mse, @mape, @smape, @theil) work as:

Code: Select all

!mystat = @theil(y_actual, y_forecast)

where y_* are series objects.

[Saakshi]

Hello Gareth, thank you for help.

I did post the command but was unable to generate the result. Following is the command file. Kindly let me know where am I making mistake.

gg_outsmpl_trial.prg

(3.23 KiB) Downloaded 647 times

Gareth, I am very new to the command program, so kindly help me out this.

Thanks,
Saakshi

[EViews Gareth]

You don't appear to be storing the theil statistic anywhere.

[Saakshi]

HORIZON h=1 h=2 h=3 h=4
Bias 0.522 1.026 1.434 1.591
MSE 3.136 5.300 8.923 12.197
RMSE 1.771 2.302 2.987 3.492
SE 1.692 2.061 2.620 3.109
MAE 1.486 2.000 2.466 2.525
MAPE 0.664 1.051 0.910 0.648
THEIL 0.000 0.000 0.000 0.000

This is my result window.

[Saakshi]

You don't appear to be storing the theil statistic anywhere.

Where should I store the statistics ideally?
Am I missing something here?

[Saakshi]

Hello, I understand this must be a very simple question but I am in the learning phase and I badly need your help.

kindly help!

[Saakshi]

All of the forecast evaluation type statistics (@rmse, @mse, @mape, @smape, @theil) work as:

Code: Select all

!mystat = @theil(y_actual, y_forecast)

where y_* are series objects.

Hi Gareth,

I am trying to write following command for Theil. Could kindly you tell me what mistake is in this?
THEIL=@sqrt(RMSE)/@sqrt({%v}-{%v}(-1))*@sqrt({%v_f}-{%v_f}(-1))

Thanks,
Saakshi

[EViews Gareth]

I see nothing wrong.