Data Visualization (AD-703 (A)) - Unit 5 Short Notes

How unit 5 is examined

This unit covers time-series, big data, text and multivariate visualization, storytelling, dashboards, ethics and case studies; the marks sit in big data, text data and Tableau dashboard creation (7 marks each, Dec 2024).

Time-Series data visualization

<span style="display:inline-block;padding:.16em .6em;border:1.5px solid currentColor;border-radius:999px;font-size:.68em;font-weight:700;letter-spacing:.06em;text-transform:uppercase;opacity:.75">Not asked since 2022</span>

Definition. <mark>Time-series visualization plots data points recorded at successive time intervals, with time on the x-axis, so that trend, seasonality and anomalies become visible.</mark>

Key points.

1. The line chart is the default choice because the connecting line shows continuity and direction over time.
2. A trend is the long-term rise or fall, seasonality is a pattern repeating at fixed intervals, and noise is random variation.
3. Area charts, stacked areas, candlestick charts and calendar heat maps suit volume, finance and daily patterns respectively.
4. Moving averages smooth the noise: $MA_t = \frac{1}{k}\sum_{i=0}^{k-1} y_{t-i}$.
5. Keep time intervals evenly spaced and mark events or forecast bands directly on the line.

Big data visualization

<span style="display:inline-block;padding:.16em .6em;border:1.5px solid currentColor;border-radius:999px;font-size:.68em;font-weight:700;letter-spacing:.06em;text-transform:uppercase;opacity:.75">Low weight</span>

Definition. <mark>Big data analysis is the process of examining datasets too large, fast or varied for traditional tools to uncover patterns and insight, and big data visualization presents those results graphically.</mark>

Key points.

1. Big data is described by the Vs: Volume (size), Velocity (speed of arrival), Variety (structured and unstructured forms), plus Veracity and Value.
2. Volume is handled by aggregating, sampling or binning, so millions of points become density plots, heat maps or hexbin charts.
3. Velocity is handled by streaming pipelines (Kafka, Spark Streaming) feeding real-time dashboards; variety by integrating tables, logs, text and images.
4. Tools include Hadoop, Spark, Tableau, Power BI, D3.js and Plotly, and they turn raw data into faster insight and better decisions.

Asked: [7 marks] (Dec 2024) What is big data analysis? Explain how it will help in data visualization.

Text data visualization

<span style="display:inline-block;padding:.16em .6em;border:1.5px solid currentColor;border-radius:999px;font-size:.68em;font-weight:700;letter-spacing:.06em;text-transform:uppercase;opacity:.75">Low weight</span>

Definition. <mark>Text visualization converts unstructured text into graphics by first cleaning and counting it, then showing frequencies, sentiment and topics.</mark>

Key points.

1. The challenge is that text is unstructured, high-dimensional and noisy, so it must be tokenized, lowercased and stripped of stop words first.
2. A word cloud sizes each word by frequency, and a frequency bar chart ranks the top words precisely.
3. Sentiment analysis scores text as positive, neutral or negative and plots it as bars or a line over time; topic models (LDA) group words into themes.
4. Example: for 1,000 product reviews, clean the text, draw a word cloud (battery, price, camera), a bar chart of top words and a sentiment pie; tools are Python (NLTK, WordCloud, Matplotlib) and Tableau.

Asked: [7 marks] (Dec 2024) How can we effectively visualize the text data? Explain with help of the example.

Multivariate data visualization

<span style="display:inline-block;padding:.16em .6em;border:1.5px solid currentColor;border-radius:999px;font-size:.68em;font-weight:700;letter-spacing:.06em;text-transform:uppercase;opacity:.75">Not asked since 2022</span>

Definition. <mark>Multivariate visualization displays three or more variables together to reveal relationships among them.</mark>

Key points.

1. Scatter plot matrices show every pairwise relationship, and parallel coordinates draw each record as a line across variable axes.
2. Bubble charts encode a third and fourth variable as size and colour; heat maps show a correlation matrix.
3. Faceting (small multiples) repeats one chart per category for easy comparison.
4. PCA or t-SNE reduces many dimensions to two so that clusters can be plotted.

Storytelling with data

<span style="display:inline-block;padding:.16em .6em;border:1.5px solid currentColor;border-radius:999px;font-size:.68em;font-weight:700;letter-spacing:.06em;text-transform:uppercase;opacity:.75">Not asked since 2022</span>

Definition. <mark>Data storytelling combines data, visuals and narrative to communicate insight so that the audience understands and acts.</mark>

Key points.

1. Know the audience and the one message (the big idea) before choosing any chart.
2. Structure the story as context, conflict (the problem the data reveals) and resolution (recommended action).
3. Remove clutter and use colour and annotation to focus attention on the key point.
4. Choose the chart that fits the message, and end with a clear call to action.

Dashboard creation

<span style="display:inline-block;padding:.16em .6em;border:1.5px solid currentColor;border-radius:999px;font-size:.68em;font-weight:700;letter-spacing:.06em;text-transform:uppercase;opacity:.75">Low weight</span>

Definition. <mark>A Tableau dashboard is a single view that combines several sheets, filters and text so users can monitor key metrics interactively.</mark>

Steps.

Step 1: Connect to the data source (Excel, CSV, SQL) and clean or join tables.
Step 2: Create individual sheets: bar, line, map, each with measures and dimensions.
Step 3: Open a New Dashboard, set the size and drag sheets onto the layout.
Step 4: Add filters, actions (highlight, filter), legends and titles for interactivity.
Step 5: Arrange with containers, test on devices, then publish to Tableau Server or Public.

Example. A sales dashboard joins a sales CSV, makes a monthly line sheet and a region map, and adds a region filter that updates both.

Asked: [7 marks] (Dec 2024) Explain step by step process of a dashboard creation in Tableau.

Ethical considerations in data visualization

<span style="display:inline-block;padding:.16em .6em;border:1.5px solid currentColor;border-radius:999px;font-size:.68em;font-weight:700;letter-spacing:.06em;text-transform:uppercase;opacity:.75">Not asked since 2022</span>

Definition. <mark>Ethical visualization means presenting data honestly, without distorting, hiding or misusing it.</mark>

Key points.

1. Truncated axes and inconsistent scales exaggerate differences and mislead.
2. Cherry-picking time ranges or data omits inconvenient facts.
3. Personal data must be anonymized to protect privacy and consent.
4. Colour choice should be accessible to colour-blind viewers, and sources and limits must be disclosed.

Case Studies for Finance-marketing, and insurance healthcare

<span style="display:inline-block;padding:.16em .6em;border:1.5px solid currentColor;border-radius:999px;font-size:.68em;font-weight:700;letter-spacing:.06em;text-transform:uppercase;opacity:.75">Not asked since 2022</span>

Definition. <mark>Case studies show how visualization solves domain problems in finance, marketing, insurance and healthcare.</mark>

Key points.

1. Finance uses candlestick charts and risk heat maps to track stock trends and portfolio risk.
2. Marketing uses funnel charts and campaign dashboards to track conversion, customer segments and ROI.
3. Insurance uses claim maps and fraud-anomaly plots to spot risky regions and suspicious claims.
4. Healthcare uses patient dashboards, epidemic maps and time series of admissions to guide treatment and resource planning.

Last-minute revision

• Time series: time on the x-axis, line chart, trend, seasonality, moving average.
• Big data Vs: Volume, Velocity, Variety, Veracity, Value.
• Big data tools: Hadoop, Spark, Tableau, Power BI, D3.js.
• Text pipeline: clean, tokenize, remove stop words, then word cloud, bar, sentiment, LDA.
• Multivariate: scatter matrix, parallel coordinates, bubble, facets, PCA.
• Storytelling: context, conflict, resolution, call to action.
• Tableau dashboard: connect, sheets, dashboard, filters, layout, publish.
• Ethics: no truncated axes, no cherry-picking, protect privacy.
• Cases: finance candlestick, marketing funnel, insurance claim map, healthcare dashboard.

Memory hooks

• Big data: "3 Vs first, then Veracity and Value".
• Story: "Context, Conflict, Resolution".
• Dashboard: "Connect, Create, Compose, Control, Combine-publish".
• Ethics: "Honest axis, honest data, honest people".

Coverage checklist

• Time-Series data visualization: definition, trend, seasonality, moving average.
• Big data visualization: Q1 (7 marks, Dec 2024).
• Text data visualization: Q3 (7 marks, Dec 2024).
• Multivariate data visualization: techniques and dimension reduction.
• Storytelling with data: structure and principles.
• Dashboard creation: Q2 (7 marks, Dec 2024).
• Ethical considerations in data visualization: misleading charts, privacy.
• Case Studies for Finance-marketing, and insurance healthcare: four domains.