Week 21 · learning day 4
ML versus deep learning
Introduction to Data Science and Analytics · 45–60 minutes
Today’s outcomes
- Explain ml versus deep learning in the context of introduction to data science and analytics.
- Design a product decision using the source outcome rather than intuition alone.
- Produce a reusable section of an analytics decision brief.
Source trace
W21-O04Differentiate between machine learning and deep learning.
Core lesson
Make the choice inspectable
ML versus deep learning matters when it changes a real allocation of attention, money, time, or delivery capacity. Analytics turns product behaviour into evidence for a specific decision.
Start by naming the decision and the uncertainty around it. Separate evidence from assumptions, compare at least one alternative, and state what would make you revise the choice.
The source outcome for today is: Differentiate between machine learning and deep learning.
What decision will ml versus deep learning improve, what evidence is sufficient for that decision, and what is the cost of being wrong?
Worked example
ML versus deep learning in practice
For a usage-insights product for account teams, the product manager must design a choice about ml versus deep learning. The team records the target user and outcome, the evidence currently available, the strongest alternative, and the next reversible test. The recommendation is written as a choice with a reason—not as a list of features.
Do the work · 20 minutes
Turn the idea into a decision
- Choose a product you know and write the specific decision that ml versus deep learning should support.
- List two pieces of evidence, two assumptions, and one credible alternative.
- Make a recommendation in three sentences and add one condition that would change it.
Save to your portfolio
An analytics decision brief — section: ML versus deep learning
Knowledge check
Answer before opening
What is the decision at the centre of ml versus deep learning?
A good answer names an accountable choice, not merely an activity or output.
How should evidence and assumptions be separated?
Label observed facts, interpretations, and untested beliefs explicitly so the next learning step is visible.
What makes the recommendation revisable?
It includes a trigger, threshold, or new evidence that would justify changing course.
What should the portfolio artefact communicate?
The context, considered alternatives, chosen direction, rationale, evidence, and remaining risk.
Spaced review
Reconnect the learning
Reflection