By the time of the deadline, each team must
(1) submit a CSV file with predicted labels to the platform, and get scored and ranked based on the public test dataset
(2) share a report/article with details using Write Article button (code included) on the Scriptedin platform, which contains the link to the team's GitHub account (see Test Procedure using the Private Dataset for additional requirements)
(3) optionally share a notebook using Share Notebook button if the team use Jupyter notebook. Currenly only Jupyter Python notebooks are supported on the platform.
Test Procedure using the Private Dataset
The test procedure is as follows:
1. Each team provides an executable file on Ubuntu 18.10 for being used with the private dataset for prediction (not training). It takes a single example (row) of test data on the command line and generates the prediction and write the output to stdout in the format of 0 1. No other character is accepted. It is up to the team how to store and use the given test data for time before t to predict at t.
2. Each team provides instruction (e.g. compiler, libraries, procedures, etc.) showing how to running their code.
3. Each team provides an introduction of the team (e.g. names of members, advisor, university or company, location, etc)
3. Each team provides a GitHub account with all the code and information above. The link to the Github account should be at the bottom of the article shared on the platform.
4. The committee downloads code for testing. Data is fed line by line from the private dataset as the following example:
If timeout or error, the current line will be skipped and the testing function will process the next line. Results will be calculated based on all the prediction outputs available