Instructions for Using Kaggle 1. Registration
The first step for using Kaggle is creating an account. To do so, you can access the Kaggle homepage and click on the register option at top right corner of the screen.
Please use the Register with Google option and use your student.unimelb.edu.au email address to make an account.
PLEASE ONLY USE YOUR STUDENT ID AS YOUR TEAM NAME.
NOTE: We will only consider submissions under the correct Student
ID. All the other submissions are considered fake and will be ignored.
If you made a mistake, you could update your TEAM NAME, in your Kaggle profile.
2. Competition
The COMP90049 2021SM2 Assignment3 is a private competition so only people who have access to this link can participate.
Link: https://www.kaggle.com/t/a3e05cc026bc420892a33b7228a3178a
After accessing the competition page, you need to “Join” the competition by clicking on the option on the top-right corner and accepting the rules.
Your prediction file needs to be in .csv format (.csv stands for Comma Separated Values file) Your CSV file should exactly two columns and 19895 rows. Your file should have a header row:
{‘tweet_id’, ‘sentiment’} and 19894 prediction rows.
After that you would be able to “Submit Predictions” using the provided option.
If your prediction file has the correct format (2 columns, 19895 rows, correct header and correct id-s) it will be loaded in Kaggle Leader Board successfully.
After a successful submission, Kaggle will give you a score (the accuracy of your test data predictions using 30% of the data). And you can also find the ranking of your results using the public leader board.
NOTE: We are checking your prediction accuracy results on 100% of the data using the private Leader Board.
It is because we do not want you to try and improve your rank just by overfitting your results for the test data (using excessive try and error submissions on Kaggle).
You can only submit up to 8 predictions on each day. It is important to keep in mind that we are NOT marking the accuracy of your model, we are assessing your ability and skills in developing and analyse of a logical argument about the problem of tweet sentiment classification, using different Machine Learning methods.
Prior to competition close, you may select a final submission out of the ones submitted previously – by default the submission with highest public leader board score is selected by Kaggle.