Submission

-> Submissions are closed <-

Submitting your Forecasts and Prediction Intervals

In order to submit your forecasts, please sign in first. Then, click on the “MyAccount” tab of the navigation bar. On your user profile, click on the settings icon and select the “Edit Profile” choice. You can now upload your submission as a .csv file under the category “Submissions”. Please note that although you can make multiple submissions, only the most recent one will be considered for evaluation.

For every time-series in the training set, the submission file should contain h+1 columns: series id and h forecasts, where h is the respective forecasting horizon. In brief, the file should have the following format:


Y1,F1,F2,...,F6
Y2,F1,F2,...,F6
...
M1,F1,F2,...,F18
...

The same format and process should be followed for submitting the Prediction Intervals. Yet, in this case two different files must be submitted, one for the upper and one for the lower bounds. There is a special form for uploading each file to the system.

Please remember to update your profile after submitting your forecasts so that these are properly saved.

The template can be found here, where Naive is used to extrapolate the M4 series. The “Submission Guide” can also be found here.

Submitting your code for reproducing the results

After submitting your forecasts, you should also submit the code used for generating and reproducing your results, as well as a short description of the method utilized. To do so, please follow the next steps:

1. Email m4@fsu.gr to request access to our GitHub Repo. Please include your GitHub username or e-mail.
2. After the M4-Team grants you access, navigate to the M4 Methods Repository in M4Competition’s account and select “Upload Files”.
3. Select the files you wish to include in your submission. The files should contain in one folder your source code and a .pdf file describing your method (see instructions ). The template for submitting the description of your method is available here. If your organization cannot fully share the source code of your solution, you may provide an executable (black box) by creating a release (see instructions ).
4. Create a new branch named after your method’s name and click “Commit Changes”.

January 12, 2018

By continuing to use the site, you agree to the use of cookies. more information

The cookie settings on this website are set to "allow cookies" to give you the best browsing experience possible. If you continue to use this website without changing your cookie settings or you click "Accept" below then you are consenting to this.

Close