Data Analytics interview exercise

We want you to create a simple machine learning model to see how you are doing with learning new things.

Allowed time

1 Week since you receive the mail

The candy problem 🍭

We want to predict the sugar level 🍧 in a particular type of candy 🍬 using a series of features corresponding to it.

Resources

This is the Dataset: https://www.kaggle.com/fivethirtyeight/the-ultimate-halloween-candy-power-ranking/

Help documentation

We leave you some documentation and tips to help you achieve this exercice how to get there:

• https://machinelearningmastery.com/linear-regression-for-machine-learning/
• https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LinearRegression.html
• https://apple.github.io/turicreate/docs/userguide/supervised-learning/regression.html
• https://realpython.com/linear-regression-in-python/
• If you have a Mac, CoreML is very easy to use to do this kind of work: https://developer.apple.com/documentation/createml/creating_a_model_from_tabular_data

You are totally free on the tools (Python + Any library, or CoreML, as you wish)

Results expectation

• We expect you to train your model on 90% of the dataset and give us your predictions for the remaining 10%.

Files expected

• A .csv file with your predictions only.
• Source code used to create the model (You can send a link to a git or Google Colab or anything else)
• A quick Readme in markdown to explain the difficulties and how you handled them.

Final touch

Document yourself long before coding and try to do what you can. If you can’t predict but you have something that should work, send us your code with explanations anyway.

This test is not eliminatory, it allows us to see if you can manage alone with a rather complicated subject.