Hello!
I am working currently on my dissertation and a part of it concerns some stats issues that I am not familiar with.
Specifically I need to perform Time Series Forecasting and I am considering using SPSS. Actually I have installed SPSS Statistics 17.
The data I am using is ratings for a movie over time, comes from xls or csv files and includes two columns: one with dates, and one with movie ratings.
A detail about dates is that they are not continuous and the records of my data set are about 1500. More specifically the data is like that:
04/04/2004 5
05/04/2004 4.5
07/04/2004 2
10/04/2004 3.3333
.
.
.
So from these 1500 pairs of values(date,rating) I would like to use 1200 as training set, and based on it, to predict ratings for the remaining.
I select the cases that I want in Data->Select cases option, and then I click on Analyze->Forecasting->Create Models.
When the Time Series Modeler window appears, I leave as method Expert Modeler, and from other tabs I select the forecast values to appear on my graph.
The result I get regarding forecast values, is a straight horizontal line.
Do I miss something in the whole process? Is this the right approach concerning time series forecasting or is there anything else that I should use?
I have no idea about SPSS and what I should use, so please help! (In fact I didn't even know about SPSS, and now my time constraints are extremely limited).
Thanks in advance and sorry for the long post.