VCE Applied Computing: Data Analytics (Year 12) · Unit 3 — Data Analytics: Analysis and Design · 16m read

Data Analytics and Visualisation

Ever wondered how Netflix knows what you want to watch, or how companies spot trends before they happen? That's data analytics — turning messy numbers into clear, meaningful stories through smart techniques and visual displays. You'll learn to build dashboards that actually communicate insights, spot patterns others miss, and recognise when data presentations might be misleading you. This skill is gold for any career involving decision-making, from business to health to engineering.

Learning objective: Apply advanced data analytics techniques and create effective visualisations to communicate insights.

By the end of this lesson you can:

I can apply data manipulation techniques including filtering, sorting, grouping, and pivot tables.
I can design multi-chart dashboards that tell a coherent data story.
I can critically evaluate data visualisations for accuracy, clarity, and potential bias.

Data analytics turns a torch on hidden patterns. Collect, clean, analyse and visualise — and patterns step out of the dark.

Advanced data analytics and visualisation

### Data manipulation techniques

Filtering: Selecting subsets of data based on criteria. "Show only sales > $1000" or "Show only 2024 data."

Sorting: Ordering data by one or more columns. Ascending/descending. Multi-level sorts: sort by state, then by revenue within each state.

Grouping and aggregation: Combining rows that share a value and applying a function (SUM, COUNT, AVERAGE). "Total sales by region" or "Average score by subject."

Pivot tables: Powerful tool for multi-dimensional analysis. Reorganise data to view it from different angles — rows become columns, values are aggregated. Essential for exploring large datasets.

Calculated fields: Creating new data from existing columns. Profit = Revenue − Cost. Growth rate = (New − Old) / Old × 100%.

### Dashboard design

A dashboard combines multiple visualisations on one page to provide a comprehensive view of data.

Design principles:
- Hierarchy: Most important metric at top/largest. Supporting detail below.
- Consistency: Same colours, fonts, scales across charts. Blue = Category A everywhere.
- Context: Include comparison data (last year, target, benchmark) so users can judge performance.
- Interactivity: Filters and drill-downs let users explore. Click a region → all charts update.

Common dashboard layouts:
- KPI tiles across the top (key numbers with change indicators)
- Trend chart (line graph showing change over time)
- Comparison chart (bar chart comparing categories)
- Detail table (supporting data for investigation)

### Critical evaluation of visualisations

Misleading techniques to identify:
- Truncated Y-axis: Starting the axis above zero exaggerates differences.
- Cherry-picked timeframe: Showing only the period that supports a narrative.
- 3D charts: Distort proportions and make accurate comparison impossible.
- Dual Y-axes: Can be manipulated to suggest false correlations.
- Inappropriate chart types: Using a pie chart with 20 categories, or a line graph for unrelated categories.

Questions to ask of any visualisation:
1. What is the source of this data? Is it reliable?
2. What is not shown? What has been excluded?
3. Does the visual accurately represent the underlying data?
4. What conclusion does it guide you toward — and is that conclusion justified?

Quick reference

| Stage | Activity |
|---|---|
| Acquire | Primary/secondary data |
| Clean | Remove errors |
| Analyse | Trends, statistics |
| Visualise | Charts, infographics |
| Communicate | Insights |

Key ideas at a glance

Data integrity: Accuracy, timeliness.
Choose visualisation: Purpose.
Avoid misleading charts: Axes.
Cite sources: Credibility.

Worked example (Medium)

A school has collected student performance data across 5 subjects over 3 years. Design a dashboard to help teachers identify trends and at-risk students.

Identify key metrics. Average score by subject (trend over 3 years). Number of students below benchmark. Subject comparison (which subjects are improving/declining). Individual student tracking.
Dashboard layout. Top row: KPI tiles — overall average, % meeting benchmark, year-on-year change. Middle: line graph showing average score per subject over 3 years (trend). Bottom left: bar chart comparing current year averages across subjects. Bottom right: table of at-risk students (below benchmark in 2+ subjects).
Interactivity. Filter by year level. Click a subject → line graph highlights that subject. Click a student → shows individual performance history.
Design choices. Consistent colour for each subject across all charts. Red indicators for declining trends and below-benchmark scores. Comparison to school and state benchmarks for context.

Answer: An effective dashboard combines KPI tiles, trend charts, comparison charts, and detail tables — each serving a different analytical purpose while maintaining visual consistency.

Worked example (Hard)

Identify and explain three ways this visualisation is misleading: a bar chart with a Y-axis starting at 90% showing satisfaction scores of 91%, 93%, and 95% for three products.

Truncated Y-axis. Starting at 90% instead of 0% makes the 4-percentage-point difference between products appear enormous — visually, Product C looks ~5× taller than Product A, when the actual difference is only 4%.
Missing context. No margin of error or sample size. If surveys had 50 respondents, the difference between 91% and 95% may not be statistically significant. The chart presents random variation as meaningful difference.
Implied conclusion. The visual guides viewers toward "Product C is dramatically better" when the reality is "all three products have very high satisfaction with minor differences that may not be significant." The chart design creates a narrative the data doesn't necessarily support.

Answer: Truncated axes, missing statistical context, and visual exaggeration can make small differences appear dramatic. Always check: does the visual representation match the actual data?

Common mistakes

Using too many chart types in one dashboard This is a common error. Fix: Stick to 3-4 chart types maximum. Too much variety creates visual confusion. Each chart should serve a clear analytical purpose.
Creating charts without clear titles and labels This is a common error. Fix: Every chart needs: a descriptive title, labelled axes with units, a legend if using multiple series, and a data source reference.
Presenting raw data without aggregation This is a common error. Fix: Analytics means processing data into insights. A table of 10,000 rows is data, not analysis. Summarise, aggregate, and visualise to create meaning.

Data storytelling

Data storytelling is the practice of building a narrative around data insights — guiding the audience from a question, through evidence, to a conclusion. The best dashboards don't just display data; they tell a story. This involves: starting with a clear question ("Are our sales growing?"), presenting evidence in logical sequence (trend → comparison → detail), and reaching a supported conclusion ("Sales grew 15% but are concentrated in one region — we need to diversify"). Data storytelling is increasingly valued in business, government, and research because data without narrative is overwhelming, and narrative without data is unsubstantiated.

Practice questions

A truncated Y-axis on a bar chart (starting above zero) is misleading because it:
1. Makes the chart harder to read
2. Visually exaggerates the differences between values
3. Changes the actual data values
4. Removes important categories
(1 marks)
Explain three principles of effective dashboard design.
(3 marks)

Summary

Data manipulation: filtering, sorting, grouping, pivot tables, calculated fields.
Dashboard design: hierarchy, consistency, context, interactivity. 3-4 chart types maximum.
Misleading charts: truncated axes, cherry-picked timeframes, 3D distortion, dual Y-axes.
Critical evaluation: check source, scale, what's excluded, and whether the visual matches the data.
Data storytelling: question → evidence → conclusion. Narrative makes data actionable.