Re: [Gretl-users] Time series prediction
Hi again,
Am I missing something? In my case the data is not actually
interpolated. I doesn't give me the "predicted" values for 55 and 56. It
only shows the fitted values for the data I provided, but not the data
I'm missing. Here is my output:
For 95% confidence intervals, t(7, 0,025) = 2,365
datavalues prediction std. error 95%
interval
1 215,00 24,31 121,471 -262,93
- 311,54
2 220,00 48,61 122,030 -239,94
- 337,17
3 200,00 72,92 122,957 -217,83
- 363,66
4 195,00 97,22 124,243 -196,56
- 391,01
5 190,00 170,14 130,132 -137,58
- 477,85
6 185,00 194,44 132,723 -119,40
- 508,29
7 170,00 218,75 135,600 -101,89
- 539,39
8 150,00 243,06 138,744 -85,02
- 571,13
Forecast evaluation statistics
Mean Error 56,944
Mean Squared Error 12871
Root Mean Squared Error 113,45
Mean Absolute Error 94,757
Mean Percentage Error 24,365
Mean Absolute Percentage Error 48,319
Theil's U 7,1394
Bias proportion, UM 0,25193
Regression proportion, UR 0,74351
Disturbance proportion, UD 0,0045524
As you can see, the forecast only shows the fitted values (8 value). It
should be 10 values with the 1955 and 1956 values. Note the my current
OLS estimation is very poor (see the bad prediction values), but that
doesn't matter now.
Peter
On 2014/05/18 12:24 PM, Narandra Dashora wrote:
Peter
I used GRETL software . First I created the sheet on Excel and
imported the file on GRETL. Then I used the window Ordinary least
Square and got the regression results. The again I used OLS and click
Forecasts . The results have been the prediction for 1955-56. . Please
feel free to communicate . I am from India therefore there may be time
lag in communication
On Sun, May 18, 2014 at 11:20 AM, Huffelpuff <huffelpuff420(a)gmail.com
<mailto:huffelpuff420@gmail.com>> wrote:
Hi,
Thanks for this Narandra. I'm still a bit confused how you
actually did the interpolation.
This is what I've done:
1. Create a dataset with 8 entries. The dataset has a variable
with the indexes (1, 2, 3, 4, 7, 8, 9, 10) and a second variable
that holds the stock prices.
2. I then estimate a model (for instance AR(1) or ARIMA) and used
the stock prices as dependent variable and the indexes as the
regressor.
3. Once estimated, I tried to forecast the values, but it only
calculates the values for the given indexes, but not for index 5
and 6.
How exactly did you get the interpolated values? Directly via the
GUI or with a hansl script?
Peter
On 2014/05/17 04:42 PM, Narandra Dashora wrote:
> Give time series a numerical value such as 1 for 1950 and so on ,
> But Give 1956 the value 7 . This will give you regression equation
> The put in the value of X. The solution of your problem may be by
> using
> Time Stock
> 1 215
> 2 220
> 3 200
> 4 195
> 7 190
> 8 185
> 9 170
> 10 150
>
>
>
>
> _______________________________________________
> Gretl-users mailing list
> Gretl-users(a)lists.wfu.edu <mailto:Gretl-users@lists.wfu.edu>
> http://lists.wfu.edu/mailman/listinfo/gretl-users
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu <mailto:Gretl-users@lists.wfu.edu>
http://lists.wfu.edu/mailman/listinfo/gretl-users
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
Attachments:
- attachment.html (text/html — 11.7 KB)