A 4.2.1 / A 4.2.2 Data cleaning and feature selection
A4.2.1/A4.2.2 Data cleaning and feature selection
A4.2.1 emphasises data cleaning, as data quality directly impacts model performance. Techniques include handling outliers, removing duplicates, correcting/filtering data, transforming formats, and managing missing values via imputation/deletion/predictive modelling. A4.2.2 focuses on feature selection to identify and retain the most informative data attributes, using strategies like filter, wrapper, and embedded methods
Learning Objectives
By the end of this lesson, students will be able to:
- Describe the significance of data cleaning and explain the impact of poor data quality on model performance.
- Identify and describe various data cleaning techniques for handling issues like missing data, duplicates, outliers, and incorrect formatting.
- Describe the purpose of normalisation and standardisation as preprocessing steps.
- Describe the role of feature selection in improving model efficiency and preventing overfitting.
- Describe the three main strategies for feature selection: filter, wrapper, and embedded methods.
Approaches to Learning (ATL) Elements
- Thinking Skills (Critical Thinking): Analysing "dirty" datasets to identify specific problems; evaluating the trade-offs between different data cleaning techniques (e.g., deletion vs. imputation).
- Problem-Solving Skills: Applying a systematic process to identify issues in a dataset and propose a logical sequence of steps to clean it.
Lesson Resources
Presentations
eBook
To make the eBook page available for students use the page manager in the student access section to make it visible.
A4.2.1 Describe the significance of data cleaning.
Other Resources
To make the Worksheets available for students use the page manager in the student access section to make it visible.
A 4.2.1 / A 4.2.2 Worksheet Answers
A4.2_Clean_the_Sheet_Teacher_Key.xlsx
Detailed Lesson Plan
| Timing (Mins) | Description of Activity | Slide # |
|---|---|---|
| 0-5 | Introduction and Hook: Start with the “Gourmet Chef” analogy. Ask students: “Can a world-class chef make a delicious meal with rotten, low-quality ingredients?” Introduce the principle of “Garbage In, Garbage Out” (GIGO) in machine learning. ATL Focus: Thinking Skills (Metacognition). | 1-2 |
| 5-20 | Lecture: The Importance of Data Cleaning (A4.2.1): Explain why data quality matters (model accuracy, fairness and reliability). Cover techniques for common data problems: missing data (deletion vs imputation with mean or median); errors (outliers, typos, invalid formats); duplicates and irrelevant data. Briefly introduce normalisation vs standardisation. | 3-6 |
| 20-35 | Practical Activity: “Clean the Sheet”, Part 1: Give students the Clean the Sheet spreadsheet (gym membership data). In pairs, students act as “Data Janitors”: find at least 8 problems, and for each one record the row, the type of problem and the action they would take, with a reason where more than one action is sensible. Take feedback on the first seven rows using slide 8, then draw out the extra cases from the teacher key: the near-duplicate, the missing target, and the genuine outlier that should be kept. ATL Focus: Problem-Solving and Communication Skills. | 7-8 |
| 35-45 | Lecture: Feature Selection, Why Less is More (A4.2.2): Explain the role of feature selection: to reduce noise, prevent overfitting and decrease training time. Describe the three strategies at a conceptual level: filter methods (selecting features before the model), wrapper methods (using the model to select features), embedded methods (the model selects features during training). | 9-12 |
| 45-55 | Practical Activity: “Clean the Sheet”, Part 2: Using the same spreadsheet, pairs decide for each column whether to keep it or drop it as an input to the model, with a one-line reason. Renewed is the target, not an input. Draw out the identifiers, the constant Currency column, the redundant Age and BirthYear pair, and the irrelevant FavouriteColour. Then ask which of the three strategies would drop each one. | 9-12 |
| 55-60 | Consolidation and Exit Check: Each student writes one cleaning problem from the sheet with the action they would take, and one column they dropped with the strategy that would remove it. Reiterate that data preprocessing comes before any model is trained. Preview the next lesson: dimensionality reduction and the “curse of dimensionality”. | 13 |
Teacher Notes: Challenges & Solutions
- Challenge: The concepts, especially the different techniques for handling missing data or the feature selection strategies, can seem abstract and dry. The extended dataset carries both halves of the lesson, so feature selection is practised on data students have already cleaned rather than taught by analogy alone.
- Solution: The practical, hands-on "Clean the Sheet" activity is critical. It makes the problems tangible and forces students to think like data scientists. For feature selection, rely heavily on the analogies provided on the slides to make the concepts intuitive.
- Challenge: Students might get bogged down in the mathematical details of statistical tests (for filter methods) or normalisation formulas.
- Solution: The syllabus requires students to describe these concepts, not implement them mathematically. Reassure them that the goal is to understand what the technique does and why it's used, not the complex underlying formulas. Keep the focus conceptual.
- Challenge: Distinguishing between Filter, Wrapper, and Embedded methods.
- Solution: Use the analogies as a constant reference point. Reinforce the key timing difference: Filter (before), Wrapper (during, via trial-and-error), Embedded (during, automatically). A quick table on the board summarising the "When" and "How" for each can be a useful visual aid.
All materials on this website are for the exclusive use of teachers and students at subscribing schools for the period of their subscription. Any unauthorised copying or posting of materials on other websites is an infringement of our copyright and could result in your account being blocked and legal action being taken against you.