Time series analysis using SPSS

Some conceptual points of knowledge I don't write here, just put an example here.

The first step:

Define the date indication amount:

Open the data file, click Data, select Define date and time, and the Define Date dialog box pops up.

The starting time in the data is the first time the cells inside the data file, my first is August 1997, each line represents the monthly sales volume, so you need to select "year, month" from the "Case is" box on the left side of the "Define Date" dialog box, and enter ' 1997 ' in the Month box on the left. 8 ', which indicates that the first case starting month is August 1997, then click Confirm, so that the SPSS data file will generate 3 new variables

Such as:

　　　　

　　

Step Two:

To understand the changing trend of time series, make a sequence list, click "Analyze", select "Time Series prediction, select" Sequence Diagram dialog box, then move the ' average ' to the ' variable ' box, ' Date_ ' to the Timeline label box, click OK. Results

　　　　

According to the analysis of sequence diagram, the fluctuation of sequence is more and more big with the season, so we choose multiplication model.

Step Three:

Click Analyze, select the time series forecast, then select seasonal decomposition, pop up the seasonal Decomposition dialog box, click OK after confirming,

　　　　

Found four more variables, err represents the error analysis; SAS represents the seasonal factor correction sequence, the seasonal factor represented by SAF, and the STC represents a long-term trend and cyclic change sequence.

We can make a sequence diagram of the new four variables, averages, and Date_, first the Err, SAS, STC, and the average and date_ as a sequence diagram, with the following effect:

Make a separate SAT and Date_ time series diagram:

　　　　

The following is the beginning of the prediction:

1. Click Analyze, select Time Series prediction, then select Create Legacy Model, and then the Time Series Modeling dialog box pops up.

2. Move the mean to the dependent variable box, then determine the middle method, select the Expert Modeler item in the drop-down list, click the Condition button on the right, and pop up the time Series Modeler: Expert Modeler Condition dialog box.

3. In the Model tab of the time Series Modeler: Expert Modeler Condition dialog box, select the all models item in the Model type box, and tick the expert modeler to consider the seasonal model check box, set, click the Continue button

4. In the Time Series Modeler dialog box, switch to the Save tab, tick the Forecast values check box, click the Browse button after the XML file in the Export model criteria box, and then set the exported model file and save path, and then click OK.

After you have finished the above steps, there will be one more column of predicted values on the original data.

　　

Before we saved the predictive model, we now use that model for predictive data.

1. Click Analyze, select Time Series prediction, and then select Apply Legacy model to bring up the Apply Model Sequence dialog box. Specific operations such as:

　　　　

The last step after finishing the above is to switch to the "save" screen, tick "predict" and click OK.

From the predicted value directly to see, we can put the predicted data and raw data together to see, but also directly do sequence diagram can be.

　　　　

In this way, a time series model is completed, and the specific prediction data can be seen in the new column of data appearing above the original data.

Insufficient:

1, found that the time series of the definition of the date no year/month/day such a model, online search, some people say with grammar can time, but is not very familiar with, did not see too understand;

2, this is simply a brief introduction of the next time series of the approximate steps, not an instance of the project. I hope we can do a complete project next time.

　　　　　　

Time series analysis using SPSS