Data Visualisation and Dashboard Reporting
Tools
Data Visualisation and Dashboard Reporting
Syllabus tag: KASNEB CPA | Advanced Level | CA34S1 Business Data Analytics
1. Why visualisation matters
Data visualisation translates complex datasets into graphical representations that humans can quickly interpret. A well-designed chart communicates a key insight in seconds; a poorly designed one obscures it. The goal is not decoration — it is clarity and insight.
2. Choosing the right chart type
| Data relationship | Recommended chart |
|---|---|
| Compare categories | Bar chart (vertical: column, horizontal: bar) |
| Show trend over time | Line chart |
| Show part-to-whole | Pie chart (limited categories) or stacked bar |
| Show distribution | Histogram (continuous data) |
| Show relationship between two variables | Scatter plot |
| Show magnitude across geography | Choropleth map |
| Show density across two dimensions | Heat map |
| Show multiple metrics at once | Dashboard with KPI tiles |
Common mistakes: using 3D charts (distorts perception); using pie charts with too many slices (> 5 becomes hard to read); starting a bar chart y-axis at a value other than zero (misleading).
3. Key visualisation principles
- Data-ink ratio (Tufte): maximise the proportion of ink used to display data vs decorative elements. Remove gridlines, backgrounds, and unnecessary labels.
- Pre-attentive attributes: properties that the human eye detects automatically — colour, size, shape, position. Use these strategically to draw attention to key insights.
- Colour use: use colour to encode meaning, not for decoration. Avoid red/green combinations (colour blindness). Use a consistent palette.
- Labels: always label axes, provide a title, cite data sources, and note the time period.
4. Dashboard design
A dashboard is a single screen displaying multiple visualisations of Key Performance Indicators (KPIs) for rapid decision-making. Design principles:
- Hierarchy: put the most important KPI first (top-left — where the eye goes first)
- Consistency: use the same colour conventions, chart styles, and terminology throughout
- Relevance: every element must earn its place — if a chart doesn't drive a decision, remove it
- Context: always show performance against a target, prior period, or benchmark — a number without context is meaningless
5. BI tools
Power BI (Microsoft): enterprise BI and analytics with strong integration with Microsoft 365 and Azure. Tableau: best-in-class visualisation, strong in ad hoc exploration. Looker (Google Cloud): cloud-native, emphasises a centralised data model. Excel: widely used for basic dashboards and ad hoc analysis. Python (Matplotlib, Seaborn, Plotly): preferred for custom analytics and automation. R (ggplot2): strong statistical visualisation capabilities.
6. OLAP and multidimensional analysis
OLAP (Online Analytical Processing) allows users to analyse multi-dimensional data interactively. Key operations: slice (filter on one dimension, e.g. sales in Kenya only); dice (filter on multiple dimensions); drill-down (navigate from summary to detail, e.g. from annual to monthly to daily); roll-up (aggregate up the hierarchy, e.g. from product to category to division); pivot (rotate dimensions to change the view perspective). OLAP cubes are optimised for analytical queries rather than transactional processing.