Hi I am using time series forecasting in SPSS to predict sales (screen shot attached). But I am getting warning "The forecast table is not created because no forecasts could be calculated."
Please suggest how to solve the problem?
Dataset: last figure is "sales" that I plan to predict, rest are time variables
2016 1 JAN 2016 2
2016 2 FEB 2016 3
2016 3 MAR 2016 0
2016 4 APR 2016 4
2016 5 MAY 2016 3
2016 6 JUN 2016 5
2016 7 JUL 2016 4
2016 8 AUG 2016 0
2016 9 SEP 2016 2
2016 10 OCT 2016 0
2016 11 NOV 2016 0
2016 12 DEC 2016 4
2017 1 JAN 2017 3
2017 2 FEB 2017 7
2017 3 MAR 2017 6
2017 4 APR 2017 8
2017 5 MAY 2017 9
2017 6 JUN 2017 4
2017 7 JUL 2017 2
2017 8 AUG 2017 3
2017 9 SEP 2017 0
2017 10 OCT 2017 0
2017 11 NOV 2017 2
2017 12 DEC 2017 3
2018 1 JAN 2018 0
2018 2 FEB 2018 2
2018 3 MAR 2018 1
2018 4 APR 2018 0
2018 5 MAY 2018 4
2018 6 JUN 2018 3
2018 7 JUL 2018 2
2018 8 AUG 2018 0
2018 9 SEP 2018 5
Hi there, Have you already solved the problem?
If not then please give some additional information regarding your problem.
for how many time periods you want to forecast the sale?
For my point of view SPSS is programmed to forecast certain amount of time period depending on the input you give (the number of data points you already have).
You can insert some blank cells in the data editor or select the option to forecast at the end of the period.
Hi Md Fuad, thanks for answer. This is entire dataset. Its a dummy dataset that i created. The problem is resolved if if redefine time variable going to define time and date option in data menu. But forecast doesnt seem realistic at all. These sales figures are number of orders received and i know it cannot be less than zeo. But after forecast the lower limit goes below zero and shows negative value. Would you suggest any good reference book for forecasting using SPSS?
Hello Dear,
Sorry I have no idea about the books on SPSS.
But I can understand your problem. I don't know which method of forecasting you are following (curve estimation, ARIMA, Exponential smoothing etc.)
First of all you never mention the time period you want to forecast. You have data set for January 2016 to September 2018 ,so then you want to forecast October 2018 to ??
For a good short run forecast on an average you need minimum 52 points. And the negative value of forecast you talking about is due to your confidence interval (90, 95, 99.) You can reduce the CI.
You have only 33 data points, so with this you can only forecast next 4 or 5 months after that periods forecast value would not be valid.
Thanks Md Fuad, I am using Expert Modeler. I will relook at data and get back.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis