Final Project Overview: MACHINE LEARNING

Outline
• Description
• Sections
• Final Deliverables
• Grading
Description
• The goal of the final project is for the student to apply the techniques they have learned
throughout the course. The topic is of the student’s choice but must be approved by the
instructor. Students should pick a topic that is of interest and there is sufficient data
available to perform modelling.
• The data should be real and available somewhere:
• You can use internal data but it is your responsibility to check with your employer
(you are likely under a non-disclosure agreement).
• Techniques applied can be supervised and/or unsupervised learning.
• DO NOT WAIT UNTIL THE LAST MINUTE.
Sections
• Introduction:
• Introduce the problem
• Describe your approach
• Data Description/Overview:
• Describe the data (observations, variables)
• Source/format
• Exploratory Data Analysis/Transformation:
• Summarize the data
• Articulate what patterns/relationships you are seeing in the data
• Clean and/or transform any observations/records
• Include a “Waterfall” table to identify the number of observations you changed or dropped
Sections (continued)
• Training/test/validation:
• Describe how you split the data (if applicable)
• Include a table that shows how your observations were split
• Modelling:
• Identify your modelling approach:
• Can be supervised and/or unsupervised learning
• Compare models: YOU MUST BE CLEAR ON HOW YOU EVALUATED YOUR MODEL TO ARRIVE AT
THE “BEST” SOLUTION
• Conclusions:
• Explain what model you chose and reason
• Appendix: CODE
Final Deliverables
• Proposal
• Project Status Update
• Presentation*:
• 15-Minutes: live attendance with slide deck
OR
• 15-Minutes: recorded video of slide deck if you cannot attend
*Duration can be less than 15-minutes if you feel like you can cover all the material in that time.
Grading Breakdown
TOTAL POINTS: 250 (approximately 1/3 of your grade)
• Proposal: 20 points
• Presentation (Project Status Update): 80 points
• Introduction
• Data Description/Overview
• Exploratory Data Analysis/Transformation
• Presentation (with Additional Sections): 150 points
• Training/test/validation
• Modelling
• Conclusion