Problem
Part I: Supervised Machine Learning
In supervised machine learning, which steps are taken to determine how well a machine learning model will predict labels in the future?
Select all that apply.
1. Determine the correct labels for each observation in a data set
2. Do not determine the correct labels for any observation in a data set
3. Train the machine learning algorithm on part of the data that is unlabeled
4. Train the machine learning algorithm on part of the data that is labeled
5. Test the machine learning algorithm on previously seen data that is unlabeled
6. Test the machine learning algorithm on previously seen data that is labeled
7. Test the machine learning algorithm on previously unseen data that is unlabeled
8. Test the machine learning algorithm on previously unseen data that is labeled
9. Measure how well the machine learning algorithm predicts labels on unseen data
10. Measure how well the machine learning algorithm predicts labels on seen data
Part II: Unsupervised Machine Learning
Imagine that you are applying an unsupervised machine learning algorithm to a data set containing 4 attributes, (i) information on the location, (ii) the length, (iii) the number of customers served, and (v) the total delivery cost for each of your company's 10,000 global delivery routes. Which insight below should you expect?
Select the best answer.
1. Explanation of how the four attributes predict whether a route fits your own definitions of a "Good Route" or a "Bad Route".
2. Grouping of the routes into previously unknown clusters with similar measures for each of the 4 attributes.
3. Coefficients that allow you to predict attribute (4) given information about attributes (1), (2), and (3).
4. The optimal sequence of customer deliveries for each of the 10,000 routes.