Data Visualization & Dashboards: Types, Tools & BI Impact

TAKEAWAYS:

• Data visualization converts complex datasets into intuitive graphs/charts (e.g., bar charts for sales trends, heatmaps for website traffic) to reveal patterns.
• Dashboards combine multiple visualizations into a single interface (e.g., Ncell’s customer churn dashboard) for real-time decision-making.
• Design principles (clarity, consistency, interactivity) determine whether a visualization misleads or informs—critical for TU’s case-study questions.
• Tools like Power BI, Tableau, and Google Data Studio automate dashboard creation, but mastery of manual techniques (e.g., Excel pivot charts) is exam-proof.
• Real-world impact: From Daraz’s inventory dashboards to NEPSE’s stock trend visualizations, BI tools drive revenue and efficiency.
• Exam focus: Trace how a given dataset (e.g., bank loan defaults) translates into a dashboard, including axis labels, color schemes, and drill-down features.

1. Why Visualize Data? The Problem with Raw Numbers

Data alone is inert. A table of 10,000 customer transactions tells you nothing about seasonal trends or fraud patterns. Visualization exploits the brain’s ability to detect patterns in shapes and colors faster than in text. For example:

• Nepali context: The NTC’s monthly internet usage report (a table) becomes actionable when plotted as a line chart showing the 2023 COVID-19 dip and post-lockdown recovery.
• Global context: YouTube’s heatmap of video watch-time by country helps creators target regions (e.g., India vs. Nepal) for ad revenue.

Caption: Why dashboards work: The brain processes visuals 60,000x faster than text (source: 3M Corporation study).

2. Core Visualization Techniques

The syllabus demands you classify and apply these. Never memorize types—master their use cases.

A. Chart Types and When to Use Them

WORKED EXAMPLE: Kathmandu Traffic Congestion Dataset: Police records of daily vehicle counts at 5 intersections (Thapathali, Kalanki, Lagankhel) over 6 months. Visualization Choice:

• Combined bar + line chart:
  • Bars: Total vehicles per intersection (static comparison).
  • Line: Hourly traffic spikes (dynamic trend). Why? Traffic engineers need to see both which roads are busiest (bars) and when congestion peaks (line). Dashboard Snippet:

flowchart TD
    A["Raw Data: 5 intersections × 6 months × 24 hours"] --> B["Step 1: Aggregate by intersection"]
    B --> C["Step 2: Plot bars for monthly averages"]
    B --> D["Step 3: Overlay line for peak hours (6–9 AM)"]
    C & D --> E["Dashboard: 'High-Risk Intersections'"]
    E --> F["Action: Add signals at Lagankhel (highest spikes)"]

B. Dashboard Design Principles

A dashboard fails if it’s cluttered or misleading. TU exams test these rules:

1. Clarity Over Creativity:
   • Use standard colors (blue for good, red for bad). Avoid rainbow gradients.
   • Label axes with units (e.g., "Sales in USD" not just "Sales").
2. Interactivity:
   • Drill-down: Click a bar in a sales chart to see regional breakdowns (e.g., Daraz’s "View Kathmandu sales").
   • Filters: Let users toggle time periods (e.g., "Show only 2023 data").
3. Consistency:
   • Same color scale across charts (e.g., green = profit, red = loss).
   • Align legends and fonts (e.g., Nabil Bank’s loan dashboard uses Arial 12pt).
4. Mobile-First:
   • 60% of BI users access dashboards on phones (Gartner). Test pinch-to-zoom on charts.

Caption: Nabil Bank’s loan approval dashboard on a smartphone, showing drill-down from "Total Loans" to "Delayed Payments by Branch."

3. Tools: From Excel to AI-Powered Dashboards

TU expects you to compare tools based on ease of use, cost, and features. Here’s the 2024 Nepal market breakdown:

REAL-WORLD CASE: Daraz’s Supplier Dashboard Problem: Daraz’s 50,000+ suppliers needed real-time order status updates. Solution: A Power BI dashboard with:

• Treemap: Supplier performance by order fulfillment rate.
• Gauge charts: On-time delivery % (target: 95%).
• Map: Geographic heatmap of delayed shipments (e.g., red zones in Far-West). Outcome: Reduced late deliveries by 30% in 6 months. Exam Question: "How would you visualize Daraz’s supplier data to identify bottlenecks?" Answer:

mindmap
  root((Daraz Supplier Dashboard))
    Treemap["Hierarchy: Supplier → Product → Delay Reason"]
    Gauge["KPI: On-Time Delivery % (95% target)"]
    Map["Geospatial: Delay Hotspots (e.g., Dhangadi)"]
    Alert["Red flags: <70% fulfillment → auto-notify supplier"]

4. Common Pitfalls and How to Avoid Them

TU exams often ask you to identify flaws in visualizations. Study these:

1. Cherry-Picking Data:
   • Bad: Showing only months where sales rose (ignoring the 3-month dip).
   • Fix: Always include a time axis (e.g., "Jan–Dec 2023").
2. Distorted Axes:
   • Bad: A bar chart with the y-axis starting at 50 (hiding small differences).
   • Fix: Start axes at 0 unless comparing ratios (e.g., market share).
3. Overlapping Data:
   • Bad: 20 lines in a scatter plot—no patterns visible.
   • Fix: Use small multiples (e.g., separate scatter plots for Kathmandu vs. Pokhara).
4. Ignoring Accessibility:
   • Bad: Red-green colorblind-unfriendly charts.
   • Fix: Use tools like ColorBrewer for safe palettes.

Caption: Left: Safe (viridis palette). Right: Unsafe (red-green) for 1 in 12 men.

5. Building a Dashboard: Step-by-Step

TU’s practical questions test this workflow. Follow these steps for any dataset:

1. Define the Goal:
   • Example: "Reduce customer churn for Ncell by 15% in Q3."
2. Select KPIs:
   • Churn rate, call drop %, customer service tickets.
3. Choose Visualizations:
   • Trend: Line chart for churn rate over time.
   • Comparison: Bar chart for churn by region (Province 1 vs. Province 5).
   • Root Cause: Scatter plot of churn vs. call drop %.
4. Design the Layout:
   • Place high-priority KPIs (e.g., churn rate) at the top.
   • Use white space—don’t cram 10 charts into one screen.
5. Add Interactivity:
   • Filter by date range (e.g., "Show only July 2024").
   • Drill-down from "Total Churn" to "Churn by Plan Type."
6. Test:
   • Ask a non-technical user (e.g., Ncell’s CEO) to interpret it in 30 seconds.

WORKED EXAMPLE: NEPSE Stock Dashboard Dataset: Daily closing prices of 20 top stocks (e.g., NABIL, NMB, CG). Dashboard:

flowchart LR
    A["Raw Data: 20 stocks × 1 year"] --> B["Step 1: Calculate 7-day moving average"]
    B --> C["Step 2: Flag stocks with >5% drop (red)"]
    C --> D["Step 3: Add volume bar chart"]
    D --> E["Final Dashboard"]
    E --> F["Alert: 'NMB down 7% → Buy signal?'"]
    E --> G["Drill-down: Click NMB → See 1-year trend"]

6. Advanced Techniques for TU’s High-Score Answers

To get 15/15 in case-study questions, incorporate these:

A. Small Multiples

• What it is: Repeated charts for subsets of data (e.g., sales by region).
• Example: Compare Pathao’s rider growth in Kathmandu, Pokhara, and Bharatpur using 3 identical line charts side by side.
• Why TU loves it: Shows trends without clutter.

B. Annotations

• What it is: Callouts explaining spikes/dips.
• Example: In a Khalti transaction chart, annotate the 2023 spike with "Dashain festival period."
• Code Snippet (Power BI):

  // Add a tooltip to a data point
  TOOLTIP [Transaction Date] + [Amount] + "Note: " & IF([Amount] > 5000, "High-value transaction")
  

C. Dynamic References

• What it is: Benchmarking against targets.
• Example: A gauge chart for Nabil Bank’s loan approval rate with:
  • Needle: Current rate (85%).
  • Arc: Target (90%) in green, warning (75%) in yellow.

D. Geospatial Visualizations

• What it is: Maps showing location-based data.
• Example: NTC’s internet speed test results as a choropleth map (darker = slower speeds).
• Tool: Use Power BI’s built-in map visual or Tableau’s Shapefiles.

7. In the Real World

Nepal’s business landscape relies on dashboards to outmaneuver competitors. Here’s how:

1. eSewa’s Fraud Detection Dashboard

   • Idea Used: Anomaly detection (scatter plots of transaction amounts vs. frequency).
   • How: Flags transactions >3 standard deviations from the mean (e.g., a single Rs. 500,000 payment).
   • Impact: Reduced fraud losses by 40% in 2023.

2. Daraz’s Inventory Turnover Dashboard

   • Idea Used: Treemap + heatmap combination.
   • How:
     • Treemap: Categories (electronics, fashion) sized by revenue.
     • Heatmap: Color-coded by stock turnover rate (red = slow-moving).
   • Impact: Cut excess inventory costs by 25% by discontinuing low-turnover SKUs.

3. Ncell’s Customer Lifetime Value (CLV) Dashboard

   • Idea Used: Waterfall chart for CLV components.
   • How: Breaks CLV into:
     • Average revenue per user (ARPU).
     • Churn rate.
     • Profit margin.
   • Impact: Identified that reducing churn by 10% increases CLV by 30%.

Caption: Real screenshot from Daraz’s supplier portal showing treemap of product categories by turnover rate.

8. Exam Tip: How to Score Full Marks

TU’s IT249 exams test application, not just theory. Follow this structure for 15/15 answers:

Part A: Short Questions (5 marks)

• Question: "Differentiate between a dashboard and a report."
• Answer:

  Dashboard	Report
  Interactive, real-time updates	Static, periodic (e.g., monthly)
  Visual (charts, gauges)	Tabular or text-heavy
  Example: Ncell’s live call drop %	Example: NTC’s annual traffic report
  Updated hourly/daily	Updated weekly/yearly

Part B: Case Study (10–15 marks)

• Question: "Design a dashboard for a hypothetical bank to monitor loan defaults. Include 3 visualizations and justify your choices."
• Answer:
  1. Visualization 1: Line Chart of default rates by loan type (personal, home, business) over 5 years.
     • Why: Shows trends (e.g., business loans spiking in 2020 due to COVID).
  2. Visualization 2: Heatmap of defaults by branch location (color-coded by default %).
     • Why: Identifies geographic risk (e.g., red zones in Terai).
  3. Visualization 3: Scatter Plot of loan amount vs. default probability.
     • Why: Reveals if larger loans are riskier.
  4. Interactivity:
     • Filter by loan tenure (1–5 years).
     • Drill-down from "Total Defaults" to "Defaulted Borrowers’ Credit Scores."
  5. Dashboard Layout:

     flowchart TD
         A["Loan Default Dashboard"] --> B["KPI: Default Rate (Target: <3%)"]
         A --> C["Trend: Line Chart by Loan Type"]
         A --> D["Risk Map: Heatmap by Branch"]
         A --> E["Root Cause: Scatter Plot of Amount vs. Risk"]
         B --> F["Alert: If rate >4%, notify risk team"]

Part C: Practical (5 marks)

• Question: "Given this table of monthly sales data, create a visualization to identify seasonal trends."
• Answer:
  • Step 1: Convert table to a line chart with:
    • X-axis: Month (Jan–Dec).
    • Y-axis: Sales (in Rs. lakhs).
  • Step 2: Add a trendline to highlight seasonality (e.g., spikes in Dashain/Teej).
  • Step 3: Use conditional formatting to color peaks (green) and troughs (red).
  • Tool: Excel pivot chart with a secondary axis for moving average.

9. Quick Revision Checklist

Before the exam, verify you can:

• Classify 5 chart types and their use cases (with Nepali examples).
• Design a dashboard for any given dataset (e.g., school attendance, hospital patient flow).
• Identify 3 flaws in a mock visualization (e.g., "This pie chart has 12 slices—hard to read").
• Name 2 tools and their pros/cons (e.g., Power BI for enterprises, Google Data Studio for free cloud dashboards).
• Explain interactivity using terms like "drill-down," "filter," and "tooltip."