Original article was published by Co-learning Lounge on Deep Learning on Medium
Submit Titanic Competition Solution on Kaggle
1. Navigate to the competition page you have participated in
2. Select the “Evaluation” tab to check the format for the submission
You can check the Metric to ensure what the competition is looking for you to prioritize.
Next, you can check the “Submission File Format” to view the prediction submission format.
Let’s take the example of the competition ‘Santander Customer Satisfaction’ and ‘Titanic Machine Learning from Disaster’.
As mentioned above, for ‘Santander Customer Satisfaction’ the submission is evaluated on ‘Area under the ROC curve’, therefore when making the model, our priority would be to fine-tune a model that gives us the best AUC score.
Similarly, for the competition ‘Titanic Machine Learning from Disaster’ our submission will be evaluated on the basis of ‘Accuracy’.
Next, we’ll perform the usual steps to build our model by following the ML Pipeline.
3. Finally, we will submit our predictions
Once our model is built, we shall submit the predictions of our model. The model will predict the values on our test dataset given alongside the train dataset in the competition.
To submit our predictions we’ll follow the steps given below. The steps performed below is an example of how the submission file is to be created in the desired format.
The above code is specific to the competition ‘Santander Customer Satisfaction’.
Similar to the above, we can create our submission file for ‘Titanic Machine Learning from Disaster’.
The above code is specific to Titanic Machine Learning from Disaster
If the kernel was made on Kaggle, upon committing all changes to the notebook, you may view it.
Go to the same competition page and click on ‘Notebooks’.
Here, under the ‘Your Work’ section, you should see your notebook.
Upon clicking your notebook, you should be taken to a page something like the below.
Click the Output section located in the lower right corner in the image above.
You should be greeted by a screen as shown below. Click the ‘Submit’ button and your submission should be submitted.
You’ll see a screen like the below.
Do I need to make my code public while submitting the solution?
You may share your code by enabling the Sharing feature and specifying if it needs to be Public or Private in your notebook. Further Collaborators can be added and comments can be disabled if required.
What if you made a custom kernel on some other location like your local machine?
In this case, you can do the following:-
First of all, create the submission file as given in the procedure above.
1/ Navigate to the competition page and select the Submit Predictions button.
2/ Click on the arrow button to Upload the submission file.
3/ You can write about your predictions and give a brief summary of your submission next and finally Submit the solution.
Congratulations, You should be able to submit your prediction successfully.
Author: Ashu Prasad