MLD101 · The Data Science Workflow and Problem Framing

Categories: AI & Automation
Wishlist Share

About Course

Learn how data science projects really run, from a vague business question to a framed problem with a clear target, decision, metric and baseline.

Explore · 4 hours · 4 lessons, a lab, an assessed assignment and a portfolio piece

Who it is for

Analysts, managers and beginners who want to start or sponsor data science work that answers real business questions.

Prerequisites

No programming is required; comfort with spreadsheets is helpful.

What you will be able to do

  • Describe each phase of the CRISP-DM workflow and what it produces
  • Translate a business question into a data science problem with a target, unit of analysis and decision
  • Choose a success metric and a simple baseline that a model must beat
  • Assess a dataset for availability, quality, bias and privacy risks before modeling

Lessons

  • Lesson 1 · The data science workflow
  • Lesson 2 · From business question to data science problem
  • Lesson 3 · Metrics, baselines and data readiness
  • Lesson 4 · Lab and assignment

Lab: Framing a customer churn problem

You will take a vague request from a small gym, "we want to know who is going to quit", and turn it into a framed problem. Using a sample membership dataset, you will define the target, check data availability and compute a simple baseline that any future model must beat.

Free tools: JupyterLite in the browser or Google Colab (free), with a sample subscription dataset

Assessed assignment

Write a two-page problem framing document for a real or realistic small business, such as a bakery forecasting daily demand or an insurance agency predicting renewals. Include the business question, the decision and who makes it, the target and unit of analysis, available data and its gaps, the success metric with a target value, a simple baseline and its measured or estimated performance, and key risks including privacy and fairness. End with a go or no-go recommendation for building a model.

Criterion Weight What excellent work looks like
Business framing 30% The decision, user and value are specific, and the question is answerable with data.
Technical definition 25% Target, unit of analysis and prediction timing are precise, with leakage risks identified.
Metric and baseline 25% The metric fits the decision costs, and the baseline is concrete and measured where data allows.
Risks and recommendation 20% Data quality, bias and privacy risks are realistic, and the go or no-go recommendation follows from the evidence.

Portfolio piece

A professional problem framing document with a baseline notebook that can open any data science project.

Self-paced, 12-month access. Questions are answered in writing through the student panel. A completion record is issued when you finish the course.

Next step

Continue with MLD102 · Python and pandas for Data Analysis · 4 hours · $28.00

Continue to the next course

Prefer the whole program? Program in Machine Learning & Applied Data Science · $378.00 (10% less than buying the courses separately)

Show More

What Will You Learn?

  • Describe each phase of the CRISP-DM workflow and what it produces
  • Translate a business question into a data science problem with a target, unit of analysis and decision
  • Choose a success metric and a simple baseline that a model must beat
  • Assess a dataset for availability, quality, bias and privacy risks before modeling

Course Content

MLD101 · The Data Science Workflow and Problem Framing (4 h)
Learn how data science projects really run, from a vague business question to a framed problem with a clear target, decision, metric and baseline.

  • Lesson 1 · The data science workflow
  • Lesson 2 · From business question to data science problem
  • Lesson 3 · Metrics, baselines and data readiness
  • Lesson 4 · Lab and assignment
  • MLD101 Assessment · The Data Science Workflow and Problem Framing
Scroll to Top