IB Computer Science HL · Year 2 · Lesson 51

Data preprocessing: cleaning, feature selection, dimensionality reduction

Preprocessing improves the quality and usefulness of training data before a model learns from it.

Paper 150 minutesA4.2.1 · A4.2.2 · A4.2.3
Use Next or the arrow keys.
Course homeIA hub2027 case study
Today’s targets

What you need to be able to do

2 / 9
  • DescribeThe significance of data cleaning
  • DescribeThe role of feature selection
  • DescribeThe importance of dimensionality reduction
A4.2.1A4.2.2A4.2.3
Paper 1 lensMatch the depth of every response to the command term. Previously learned content can move quickly, but retrieval must still be accurate.
Page 2 of 9
Retrieve

Rapid Recall Deck

3 / 9

Say the answer aloud before flipping. Mark secure knowledge quickly and spend time on the gaps.

  • Why is data cleaning important before training a model?: Poor-quality data can produce misleading patterns and reduce model accuracy, reliability, and fairness.
  • Which data-cleaning problems should you know how to address conceptually?: Outliers, duplicate data, incorrect data, irrelevant data, inconsistent formatting, and missing values.
  • How can missing data be handled?: By imputation, deleting affected records/fields when justified, or using predictive modelling to estimate missing values.
  • What is the difference between normalization and standardization in preprocessing?: Normalization rescales values to a common range; standardization rescales values around a common centre and spread, commonly using the mean and standard deviation.
  • What is feature selection?: Choosing the most informative existing attributes and removing less useful ones before modelling.
  • What are the three main feature-selection strategy families?: Filter methods, wrapper methods, and embedded methods.
  • What is the goal of dimensionality reduction?: To reduce the number of variables while preserving information relevant to the task, helping address problems associated with very high-dimensional data.
  • What problems are associated with the curse of dimensionality?: Potential overfitting, higher computation/memory needs, data sparsity, weaker distance metrics, harder visualization, and a need for more samples.
Page 3 of 9
A4.2.1 + A4.2.2 · Learn

Core knowledge and application

4 / 9
A4.2.1Describe

The significance of data cleaning

Preprocessing improves the quality and usefulness of training data before a model learns from it.

Exam moveGive a detailed, accurate account of what happens or what something is like.
Required detail 1The impact of data quality on model performance
Required detail 2Techniques for handling outliers, removing or consolidating duplicate data, identifying incorrect data, filtering irrelevant data, transforming improperly formatted data, and imputation, deletion or predictive modelling for missing data
Required detail 3Normalization and standardization as crucial preprocessing steps

Explain it without notes

Describe: The significance of data cleaning in the context of a noisy dataset being prepared before model training.

  • The impact of data quality on model performance
  • Techniques for handling outliers, removing or consolidating duplicate data, identifying incorrect data, filtering irrelevant data, transforming improperly formatted data, and imputation, deletion or predictive modelling for missing data
  • Normalization and standardization as crucial preprocessing steps
A4.2.2Describe

The role of feature selection

Preprocessing improves the quality and usefulness of training data before a model learns from it.

Exam moveGive a detailed, accurate account of what happens or what something is like.
Required detail 1Feature selection to identify and retain the most informative attributes of the data set
Required detail 2Feature selection strategies: filter methods, wrapper methods, embedded methods

Explain it without notes

Describe: The role of feature selection in the context of a noisy dataset being prepared before model training.

  • Feature selection to identify and retain the most informative attributes of the data set
  • Feature selection strategies: filter methods, wrapper methods, embedded methods
Page 4 of 9
A4.2.3 · Learn

Core knowledge and application

5 / 9
A4.2.3Describe

The importance of dimensionality reduction

Preprocessing improves the quality and usefulness of training data before a model learns from it.

Exam moveGive a detailed, accurate account of what happens or what something is like.
Required detail 1The curse of dimensionality considerations may include overfitting, computational complexity, data sparsity, the effectiveness of distance metrics, data visualization, sample size increases, memory usage
Required detail 2Dimensionality reduction of variables, while preserving the relevant aspects of the data Note: Statistical techniques such as principal component analysis (PCA) and linear discriminant analysis (LDA) are beyond the scope of this course
Assessment boundaryPCA and LDA are beyond the scope of this course.

Explain it without notes

Describe: The importance of dimensionality reduction in the context of a noisy dataset being prepared before model training.

  • The curse of dimensionality considerations may include overfitting, computational complexity, data sparsity, the effectiveness of distance metrics, data visualization, sample size increases, memory usage
  • Dimensionality reduction of variables, while preserving the relevant aspects of the data Note: Statistical techniques such as principal component analysis (PCA) and linear discriminant analysis (LDA) are beyond the scope of this course
Page 5 of 9
Apply

Transfer to a new scenario

6 / 9
ScenarioA noisy dataset being prepared before model training needs a design or technical decision related to today’s topic. Explain what matters and why.
  • The impact of data quality on model performance
  • Techniques for handling outliers, removing or consolidating duplicate data, identifying incorrect data, filtering irrelevant data, transforming improperly formatted data, and imputation, deletion or predictive modelling for missing data
  • Feature selection to identify and retain the most informative attributes of the data set
  • Feature selection strategies: filter methods, wrapper methods, embedded methods
  • The curse of dimensionality considerations may include overfitting, computational complexity, data sparsity, the effectiveness of distance metrics, data visualization, sample size increases, memory usage
  • Dimensionality reduction of variables, while preserving the relevant aspects of the data Note: Statistical techniques such as principal component analysis (PCA) and linear discriminant analysis (LDA) are beyond the scope of this course
Page 6 of 9
Exam lens

Paper 1 practice

7 / 9
Build the response before checking notesUse precise terminology and match the required depth.
  1. Describe: The significance of data cleaning in the context of a noisy dataset being prepared before model training.
  2. Describe: The role of feature selection in the context of a noisy dataset being prepared before model training.
  3. Describe: The importance of dimensionality reduction in the context of a noisy dataset being prepared before model training.
Self-checkAnswer the exact command term. For explain, include mechanism/reason; for compare, pair criteria; for discuss/evaluate/justify, build supported reasoning and a conclusion.
Page 7 of 9
2027 Case Study

Generative AI research checkpoint

8 / 9

Build a reusable evidence bank for Paper 1 Section B. Keep claims technical, specific and supported.

Character consistencyKeeping a recurring person or visual subject recognizably consistent across generated images.
Know and apply in case-study context
Embedding-based approachRepresenting concepts or visual features as numerical vectors so similarity or identity can be controlled.
Know and apply in case-study context
Segmentation mapAn image-like representation in which regions are labelled by category, often used as a condition for generation.
Know and apply in case-study context
DiscriminatorThe GAN network that estimates whether an example is real or generated.
Deep for HL where tied to GAN/hybrid challenges
Mode collapseA GAN failure mode in which the generator produces too little variety and repeatedly outputs similar samples.
Deep for HL where tied to GAN/hybrid challenges
Page 8 of 9
Homework

Finish the learning cycle

9 / 9

Exam preparation — main task

Complete targeted 2027 case-study research and cumulative Paper 1/Paper 2 practice. Record evidence and technical vocabulary you can use in extended responses.

Retrieval

Repeat today’s recall deck and revisit any item marked ‘Review again’.

Designed by David Xu
© 2026 David Xu. All rights reserved.
Educators and students are welcome to use these materials for non-commercial teaching and learning with attribution. Please share the original link when possible. Reposting, redistributing modified copies, removing attribution, or commercial use requires prior permission.
End of Lesson 51.