Framing a Data Science Problem
About
An overview of key steps and considerations when framing a data science problem. It is designed as a deep dive into the process of framing a data science problem, one of the stages of the data science lifecycle.
It includes a case study where you will be able to apply the concepts and steps learned in this module.
Learning Objectives
You will be able to:
- Identify and frame a real-world business problem suitable for data science solutions.
- Identify key stakeholders and understand their perspectives and needs.
- Formulate testable hypotheses and define appropriate evaluation methods.
- Formulate clear success criteria for your project, linked to business objectives
- Understand the ethical considerations in data science projects.
- Evaluate the trade-offs involved in a data science project by assessing the required resources and comparing them to the projected return on investment (ROI)
Content
| Lesson | Est. Delivery Time | Skills |
|---|---|---|
| Slides | 60 min | • Defining the Business Problem • Understanding your Stakeholders • Clarifying Success Metrics & Hypotheses • Assessing Feasibility • Considering Ethical, Legal and Strategic Implications |
| Calculating ROI Exercise | 15 min | • Calculating ROI for Wind Farm Predictive Maintenance Project |
| Total content | 90 min | Including breaks |
Prerequisites
- Understanding of Data Science Lifecycle: Positioning problem framing within the broader data science lifecycle
- Basic understanding of Python and Jupyter Notebooks
- Familiarity with Data Science Problem Types: Understanding the distinctions between classification, regression, clustering, and time-series forecasting.
- Basic Understanding of Machine Learning Algorithms: A high-level understanding of common algorithms for each problem type (e.g., Logistic Regression, Random Forest, Linear Regression, K-Means) is helpful. They don’t need deep technical knowledge, but should know which algorithms generally apply to which problem types.
- Data Preparation Awareness: Learners should know the importance of data cleaning, transformation, and handling missing values
- Metrics for Model Evaluation: Learners should be familiar with basic evaluation metrics (e.g., accuracy, precision, recall)