Data Intelligence and Big Data · Term 2

Machine Learning and Predictive Modeling

An introduction to how algorithms learn from data to make predictions or classifications.

About This Topic

Machine learning and predictive modeling teach students how algorithms process training data to identify patterns and generate predictions or classifications. In Year 10 Technologies, focus on supervised learning where labeled examples guide the model, such as categorizing emails as spam or predicting house prices from features. Students connect this to big data applications in recommendation systems, healthcare diagnostics, and social media feeds.

Aligned with AC9DT10K01 and AC9DT10P02, this topic prompts critical examination of biased training data causing unfair outcomes, limitations of historical patterns for future forecasts, and whether algorithms can handle ethics without human oversight. These key questions build skills in data evaluation, model testing, and ethical reasoning essential for data intelligence.

Active learning excels in this abstract area. When students collect data, train simple models using tools like Teachable Machine, and test predictions in groups, they witness bias effects firsthand and debate real scenarios. This hands-on approach makes statistical concepts concrete, boosts problem-solving confidence, and encourages collaborative critique of AI systems.

Learning Objectives

• Explain how algorithms learn patterns from labeled data in supervised machine learning.
• Analyze the impact of biased training data on algorithmic fairness and outcomes.
• Evaluate the limitations of using historical data for predicting future events.
• Critique the ethical considerations of algorithmic decision-making without human intervention.
• Design a simple predictive model using a tool like Teachable Machine to classify data.

Before You Start

Data Representation and Organization

Why: Students need to understand how data is structured and organized into tables or lists to grasp how algorithms process it.

Introduction to Digital Systems

Why: A basic understanding of how computers process information is helpful for comprehending the mechanics of algorithms.

Key Vocabulary

Algorithm	A set of rules or instructions followed by a computer to solve a problem or perform a task, such as making a prediction.
Training Data	The dataset used to teach a machine learning model. The quality and characteristics of this data directly influence the model's performance and fairness.
Supervised Learning	A type of machine learning where the algorithm learns from a dataset that includes both input features and corresponding correct outputs (labels).
Prediction	An output generated by a machine learning model based on patterns learned from data, forecasting a future outcome or classifying an input.
Bias (Algorithmic)	Systematic and repeatable errors in a computer system that create unfair outcomes, often stemming from biased training data or flawed algorithm design.

Active Learning Ideas

Pairs Activity: Bias Detection in Datasets

Provide pairs with a sample hiring dataset showing gender bias in promotions. Students chart patterns, calculate error rates for subgroups, and propose balanced alternatives. Pairs present findings to the class.

⏱ 30 min·Pairs

Generate mission

Small Groups: Train Your Own Classifier

Using Teachable Machine or Scratch extensions, groups gather images or text data for categories like fruits or sentiments. They train models, test on new data, and log accuracy drops from poor training sets. Groups swap models to evaluate.

⏱ 45 min·Small Groups

Generate mission

Whole Class: Ethical Algorithm Debate

Divide class into teams to argue for or against 'Algorithms can make ethical decisions alone.' Prep with case studies like facial recognition errors, then debate with evidence from prior activities. Vote and reflect.

⏱ 35 min·Whole Class

Generate mission

Individual: Prediction Journal

Students select a personal dataset like sports scores, build a simple prediction rule, test on new data, and journal limitations like overfitting. Share key insights in a class gallery walk.

⏱ 25 min·Individual

Generate mission

Real-World Connections

Social media platforms like TikTok and Instagram use predictive modeling to recommend content to users based on their past viewing habits, aiming to keep them engaged.

Financial institutions employ machine learning to detect fraudulent transactions by analyzing patterns in spending data, flagging unusual activity for review.

Healthcare providers are exploring predictive models to identify patients at high risk for certain diseases based on their medical history and genetic information.

Watch Out for These Misconceptions

Common Misconception

Machine learning works exactly like human learning.

What to Teach Instead

Algorithms optimize math functions on data patterns, lacking human comprehension or creativity. In training activities with visual tools, students see models fail on edge cases, clarifying statistical nature. Group testing reinforces this distinction over rote explanations.

Common Misconception

Algorithms are neutral if trained on data.

What to Teach Instead

Data embeds societal biases, leading models to perpetuate discrimination. Pairs analyzing loan datasets spot skewed predictions, then balance data to compare outcomes. This active step reveals fairness needs better than theory alone.

Common Misconception

Predictions improve endlessly with more data.

What to Teach Instead

Quality matters over quantity; noisy data worsens models. Holdout testing in group challenges quantifies this, as students measure accuracy drops. Discussions connect to real limits like historical data flaws.

Assessment Ideas

Discussion Prompt

Pose the question: 'Imagine a hiring algorithm that consistently favors male candidates. What steps could a developer take to identify and correct the bias in the training data or the algorithm itself?' Facilitate a class discussion on potential solutions.

Exit Ticket

Ask students to write down one example of a prediction an algorithm might make (e.g., predicting movie preferences). Then, have them list one potential limitation or ethical concern related to that prediction.

Quick Check

Present students with two short descriptions of datasets: one that is diverse and representative, and another that is skewed towards a particular demographic. Ask them to identify which dataset is more likely to lead to a fair predictive model and explain why.

Frequently Asked Questions

How to teach bias in machine learning to Year 10 students?

Start with relatable datasets like job applications or music recommendations showing skewed outcomes. Have students in pairs visualize patterns with charts, train biased models, then retrain with balanced data. This reveals discriminatory effects concretely. Follow with class talks on real impacts, like in criminal justice algorithms, to build ethical data practices. Emphasize diverse data collection as a solution.

What limits historical data in predictive modeling?

Historical data captures past biases and rare events poorly, failing to predict shifts like market changes or pandemics. Students explore this by extending sports datasets with anomalies, testing model breakdowns. Activities highlight extrapolation risks, teaching validation techniques and the need for ongoing model updates in dynamic contexts.

How can active learning help students understand machine learning?

Active methods like hands-on model training with Teachable Machine let students input data, observe learning iterations, and test failures directly. Group critiques of biased results make abstract stats tangible. Debates on ethics engage critical thinking. This builds deeper retention and skills versus passive lectures, as students own discoveries and connect to curriculum standards.

Can algorithms make ethical decisions without humans?

Algorithms follow programmed rules and data patterns, unable to weigh moral nuances like context or intent. Classroom debates with cases like autonomous vehicles show gaps. Students evaluate by simulating decisions in groups, concluding human oversight is vital for accountability. This fosters responsible AI perspectives aligned with ethical computing standards.