This tip is part of Learn by Example using SAS® Enterprise Miner™ series which offers various SAS Enterprise Miner templates to perform commonly used data mining tasks. These tips can be especially useful for those who do not have much experience using SAS EM to help them jump start using the software.Today I would like to cover the time series example which compares the sales of five products in five states. Below is the related EM flow.
The TS Data Preparation node prepares the data for time series analysis by creating 25 time series variables where each variable corresponds to a sales series for each product and state combination.
TS Similarity node compares each time series to a specified target series. In the first TS Similarity node (TS Similarity General), no series is specified as target, so it performs pairwise comparisons among all of the 25 input time series variables by producing various similarity statistics and plots including cluster dendograms, cluster constellation plots, and distance maps.
The second TS Similarity node (TS Similarity Comparisons) performs the comparison based on a specified target series. Specifying a target series requires changes in the role of input variables, which is specified through the Metadata node.
If you would like to see the details of this analysis and view the nice similarity plots in SAS EM, go to our github repository, import the diagram (XML file) to your SAS Enterprise Miner project, and run the flow. More information about the example can be found in the PDF file.