john-on-line27 opened a new pull request, #2629: URL: https://github.com/apache/systemds/pull/2629
The main task of this pull request was to prove the correctness of this code, and this entailed building out a test suite. Since there is no inbuilt function for double and triple exponential smoothing (DES and TES), the best way to validate the algorithm was to create a reference output from the built R functions. However in doing this, two issues were found. For the double exponential smoothing, the there was an issue with the offset of the function, creating more predictions than the R function with the same amount of data, so this issue was remedied by changing the amount of data being added to the predictions array. The second issue was that the TES function would not run if the frequency was set to an odd number. This was causing an indexing exception in one of the if statements, so the range within the if function was updated to fix this issue. Additionally, other things have been added to the EMAUtils.java file, more specifically, exception/error handling, as well as comments. This should lead to an easier time for future maintainers of this code. The shortcomings to this code include the small sample size for the testing data. This was done so that the program does not run too slowly, but if more data is required to assure quality of the code, that is also something that I am willing to include. Assisted by: AI AI contributions include the generation of test data, as well as debugging the connection between the JUnit tests and the R tests. The specific AI used was ChatGPT, by OpenAI, and the specific model was GPT-6.1 Sol. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
