Skip to main content

Skills application

Instruction and application
Complete

Applying mathematical principles to ML model selection​

In this skills application, you will analyse the key mathematical considerations for selecting the most appropriate model for a specific business problem.

Skills application illustration

Context​

You've been hired as an ML engineer at MultiVision Analytics. Your team has been tasked with developing a predictive model for an e-commerce client who wants to optimise their marketing strategy by predictingCustomer Lifetime Value (CLV).

The project is currently in the model development phase of the ML lifecycle. Your manager has asked you to analyse the key mathematical considerations for selecting the most appropriate model for this problem.

Company brief: StyleVerse​

StyleVerse is a rapidly growing e-commerce platform specialising in fashion and lifestyle products. Founded in 2018, it has expanded to over 500,000 registered users.

Business challenge: Despite its growth, StyleVerse faces increasing acquisition costs. The marketing team wants to transition from a one-size-fits-all approach to a data-driven strategy that prioritises high-value customers.Project goals:

  • Accurately predict CLV for new and existing customers.
  • Identify key factors influencing CLV.
  • Enable personalised marketing based on predicted value.
  • Optimise marketing budget allocation.

Available data:

  • 50,000 customer records with complete purchase histories.
  • 15 features (demographics, purchase behaviour, engagement metrics, returns).
  • 18 months of historical data.
  • Target variable: Actual CLV.

Instructions and materials​

To successfully complete the skills application, you must:

Identify two potential models.​

  • Identify two potential regression models.
  • Explain which basic mathematical concepts (linear algebra, statistics, optimisation) are most relevant.
  • Explain how these affect the model's ability to handle the dataset characteristics.

Complete a trade-offs analysis.​

Compare the two models in terms of model complexity, computational requirements, and interpretability.

Provide a recommendation.​

  • Recommend the best solution.
  • Explain appropriate evaluation metrics.
  • Describe how it aligns with client requirements.

Tip​

Remember that model development is just one phase of the ML lifecycle. Your analysis should consider how mathematical foundations in this phase connect to the overall ML process.

Task 1: Identify Models
Edit
Identify two potential regression models and explain relevant mathematical concepts.
Your answer here...
Task 2: Trade-offs
Edit
Compare the two models in terms of complexity, computation, and interpretability.
Your answer here...
Task 3: Recommendation
Edit
Recommend a model and explain the evaluation metrics.
Your answer here...