Here learners have the opportunity to dig into a real-world data problem with all the information you need to on this session's Capstone Challenge.
Frequently Asked Questions
I signed up but haven't received a confirmation email
E-mails can take a few minutes to process. If you haven't received your confirmation email:
- Check spam folder.
- Wait up to 30 minutes to see if email arrives.
- If e-mail has not arrived in 30 minutes, contact us.
I forgot my password
There is a self-service "Forgot password?" link on the sign-in dialog. The emails with confirmation links for password resets may take a few minutes to arrive.
How do I make a submission?
In order to submit to the competition, you need to build a model that makes predictions for the test set in
the competition, and write those predictions to a
.csv file in exactly the same format as the
However, if you want to just verify that everything works, you can simply submit the submission format file,
0s for every prediction. To do that, follow these steps:
- Step 0: Download the data from the "Data Download" page
- Step 1: Go to the "Submissions" page
- Step 2: Click the "Make a new submssion" button, and then choose the
submission_format.csvfile that you downloaded from the data download page.
- Step 3: Click "Submit" score
You won't have a good score at this point, but you will at least have made one submission! (Note: you are only allowed to make 3 submissions per day.)
At this point you can to submit your token to EdX by following the instructions in the FAQ. Though, you won't have a very good grade yet with this score!
My submission is still pending/shows loading image/hasn't been scored
Occasionally, submissions can "hang" in an unscored status, meaning that you have a submission counted against your daily limit but haven't received a score yet. If this happens, please contact us and a site admin can requeue the submission for scoring.
It says something's wrong with my submission but I can't figure out what
When you are submitting an incorrect format, we can often provide helpful error messages if the error falls into some common categories (e.g. lack of headers, wrong data types, etc.). However, if you are submitting with an unusual error, you will get a generic message. Most often if you ask on the edX Discussion Board for this course, other users can help figure out what is wrong (note: do not share your whole submission file; just a sample will do). If you still can't figure out the problem or feel time pressure, please contact us.
How do I turn my score here into a grade on edX?
You will get credit on edX by pasting an encrypted token from DrivenData into a simple graded assignment form on the edX course. Here are the steps:
- Step 0: Navigate to the export page.
- Step 1: Copy the encrypted token representing your best private score (your predictions
on the holdout data):
Navigate to the graded assignment on the edX course and paste in your token:
- Step 3: Click the button to submit and see your score.
Why is my grade on EdX lower than I expected?
Remember: the leaderboard scores and leaderboard benchmarks on the competition platform show your public score, while your grade is based on your private score. In other words, you are being graded on how close your predictions were to the holdout data for which you know the inputs but not the outputs. If your grade doesn't match what you expected, make sure you are using appropriate cross validation techniques to avoid overfitting the training data!
The grading cutoffs are explained under the "Get Your Grade" widget on EdX
I'm having issues transferring my score to edX/viewing my score on edX
If there are any issues accessing this score or concerns that the score is not correct for some reason, please contact us and we will make sure everything is working properly.
I have a question about the data or how to solve the problem
In general, most questions about the dataset, methods being used in the competition, or any questions about the underlying challenge setup should be posted to the edX course Discussion Board. This way, all other users can benefit from the answers and the question asker does not receive unfair assistance. Examples of such questions are:
- I noticed X about the data — what does this mean?
- I'm having trouble getting model Y to give good predictions for this problem; what do you recommend?
- I keep trying to submit a file like Z but it's telling me the format isn't correct — what am I doing wrong?
For questions that may indicate any problems with the predictive challenge or otherwise seem worthy of escalation, please contact us.