Geographic and Spatial Visualization

Introduction

Geographic and spatial visualization uses location to reveal patterns that may be hidden in ordinary tables and charts. In Power BI, a sales analyst may compare revenue by country, a logistics manager may examine delivery density, and a public-sector team may identify service gaps by district. Maps are therefore not merely decorative reporting elements: they are analytical instruments that connect business measures with place, distance, direction, territory, and spatial concentration.

However, maps can also mislead. A location may be assigned incorrectly because a place name is ambiguous; a large region may appear important simply because it occupies more visual area; and totals may be compared without considering population, area, exposure, or opportunity. Effective geographic visualization requires both technical competence and critical judgement.

This chapter explains how to prepare location data, construct geographic hierarchies, use coordinates, select suitable Power BI map visuals, work with Azure Maps capabilities, apply layers, and communicate spatial results responsibly.

Learning Objectives

By the end of this chapter, you should be able to:

1. Explain the role of geographic fields, coordinates, and spatial hierarchies in Power BI.
2. Prepare and categorize location data for reliable geocoding.
3. Select appropriate map visuals for points, regions, routes, density, and flows.
4. Use Azure Maps layers and analytical settings appropriately.
5. Build a practical geographic report using filters, tooltips, and drill-down.
6. Identify misleading spatial comparisons, privacy risks, and governance requirements.
7. Evaluate a map critically and communicate findings without overstating evidence.

1. The Nature of Geographic Data

Geographic data combines a spatial reference with attributes. The spatial reference identifies where an observation is located; attributes describe what is measured there.

For example, a retail dataset might contain:

A location field may be categorical, such as “Nairobi,” or numeric, such as latitude and longitude. The two forms are related but not interchangeable. A city name must be interpreted by a geocoding service, whereas a coordinate pair directly identifies a position on the Earth, subject to accuracy and coordinate-system assumptions.

1.1 Geographic dimensions and measures

A geographic dimension describes place: country, state, province, county, postal code, city, store, or service area. A measure describes a quantity associated with that place: revenue, population, incidents, customers, distance, or delivery time.

The same measure may lead to different conclusions depending on its denominator. Total incidents show workload, while incidents per 100,000 residents show relative incidence. In general:

Here, Count is the observed event total, Relevant population or exposure is the appropriate denominator, and N is a reporting base such as 1,000 or 100,000. The numerator and denominator must refer to the same geography and time period. A rate is meaningful only when the denominator is conceptually related to the event.

1.2 Point, line, and polygon data

Spatial data is commonly represented in three forms:

• Points: stores, customers, hospitals, incidents, or delivery stops.
• Lines: roads, routes, pipelines, rivers, or movement paths.
• Polygons: countries, sales territories, postal zones, districts, or catchment areas.

Power BI reports often use point maps for operational analysis and filled or boundary maps for territorial comparison. A point map answers “where are the observations?” A polygon map answers “how does a measure vary across areas?”

2. Preparing Location Fields in Power BI

A map is only as reliable as its location data. Preparation should occur before visual design.

2.1 Assign data categories

Power BI uses data categories to interpret fields. Text columns should be categorized appropriately—for example, Country/Region, State or Province, City, Postal Code, Address, or Place. Latitude and longitude fields should be categorized as Latitude and Longitude and treated as numeric values.

Categorization is a semantic instruction, not a guarantee of correctness. A column called “Region” may refer to a country subdivision, a sales division, or a commercial territory. The analyst must identify the actual meaning before assigning a category.

2.2 Reduce ambiguity through geographic context

Names such as “Springfield,” “Victoria,” or “Cambridge” can refer to multiple places. Include higher-level context whenever possible:

• City plus state or province
• State or province plus country
• Postal code plus country
• Store identifier plus coordinates

A geographic hierarchy can therefore be structured as:

Continent → Country → State/Province → City → Site

The hierarchy should reflect the organization’s analytical purpose. A logistics report may need depot → route → stop, while a national sales report may need country → region → branch.

2.3 Prefer coordinates when precision matters

Latitude and longitude reduce ambiguity and are especially valuable for store, customer, asset, and incident data. Coordinates are normally expressed in decimal degrees. Latitude ranges from −90° to +90°, and longitude ranges from −180° to +180°. Positive latitude generally indicates north, negative latitude south; positive longitude generally indicates east and negative longitude west, depending on the adopted convention.

Coordinates should be checked for:

• Missing values
• Reversed latitude and longitude
• Values outside valid ranges
• Excessive rounding
• Locations placed at headquarters rather than actual sites
• Inconsistent coordinate reference systems

If a coordinate pair is reversed, a valid-looking point may appear in a completely wrong location. Validation should include plotting a sample and comparing it with known addresses.

2.4 Aggregation and duplicate points

Multiple records may share one location. Power BI may aggregate a measure by location, but the result depends on the fields placed in the visual. Distinct stores, orders, and customers must not be confused. A map showing 10,000 transactions may contain only 20 stores.

Useful measures include:

• Total revenue
• Distinct count of stores
• Distinct count of customers
• Average order value
• Median delivery time
• Revenue per square kilometre
• Incidents per 100,000 residents

The correct aggregation depends on the question. Average delivery time is not equivalent to total delivery time, and a sum of percentages is usually not meaningful.

3. Geographic Hierarchies and Drill-Down

A hierarchy allows users to move from broad geography to detailed geography. For example, a visual may begin with country-level revenue and allow drill-down to state, city, and store.

Hierarchies support two analytical movements:

1. Drill-down: moving from a higher level to a lower level, such as country to city.
2. Drill-up: returning to a broader level to observe an overall pattern.

A hierarchy should be logically nested. If cities are placed under countries without states, the structure may be acceptable for a global overview, but it may be inadequate where city names repeat across states.

Drill-down does not automatically provide causal explanation. A region with low sales may contain fewer stores, lower population, stronger competition, or incomplete data. The hierarchy helps locate the pattern; other fields are needed to interpret it.

Small multiples and decomposition

Small multiples can display the same geographic or performance measure across categories, such as product families or customer segments. This supports comparison while preserving a consistent visual scale. The decomposition tree can also help analyze a measure by geographic dimensions, although it is not itself a map.

4. Choosing a Map Visual

Power BI provides several approaches to geographic visualization. The exact visual availability and capabilities may change with product updates, tenant settings, and organizational policy, so users should verify the current environment.

4.1 Bubble or point maps

Point size can encode a measure such as revenue, while colour can encode category or performance. This is effective for locating branches or assets. Size should be used cautiously: viewers may interpret circle area rather than radius, and large symbols can overlap.

Tooltips should include the location name, measure, comparison benchmark, and relevant denominator. A tooltip such as “Revenue: $2.4 million; stores: 3; revenue per store: $800,000” is more informative than revenue alone.

4.2 Filled maps and choropleths

A choropleth shades geographic areas according to a value. It is appropriate when the variable belongs to the entire area, such as unemployment rate by district. It is less appropriate for showing the location of individual stores, because the polygon is not the store.

Choropleths should generally use normalized values—rates, percentages, or per-capita measures—when region populations or sizes differ substantially. Classification methods such as equal intervals, quantiles, or natural breaks affect the visual message. The legend should explain the classes, and the analyst should avoid excessive categories that create false precision.

4.3 Shape maps

Shape maps can display custom boundaries such as sales territories, campuses, or electoral districts. They are useful when standard administrative boundaries do not match the organization’s decision structure. Custom shapes require reliable identifiers and boundary files, along with permission to use the underlying data.

5. Azure Maps in Power BI

Azure Maps is a location-focused visualization option that can combine business data with geographic context. Depending on the configuration and available features, it may support map styles, point layers, reference layers, heat maps, and other spatial representations.

A typical Azure Maps workflow is:

1. Add the Azure Maps visual to the report.
2. Place latitude and longitude fields in the corresponding roles, or provide a well-categorized geographic field.
3. Add a measure such as sales, incidents, or volume.
4. Configure the map view and zoom level.
5. Select an appropriate map style.
6. Configure layers such as bubbles, heat, reference boundaries, or labels where available.
7. Add legends, tooltips, filters, and accessibility descriptions.
8. Test the visual with known locations and representative filter selections.

5.1 Layers

Layers separate different spatial meanings. For example:

• A bubble layer can show branch locations and sales.
• A heat layer can show order concentration.
• A reference layer can show territories, roads, or administrative boundaries.
• A label layer can identify cities or selected sites.

Layers should not compete visually. If a heat layer is opaque and strong, it may conceal individual stores. If boundaries are too prominent, they may distract from the measure. The design should establish a visual hierarchy: the analytical layer should be most noticeable, while contextual layers remain supportive.

5.2 Map style and basemap context

A road map, light map, aerial image, or dark map serves different purposes. A muted basemap usually improves the visibility of analytical marks. Aerial imagery may help with site context but can introduce visual noise. Map labels should be legible at the intended zoom level.

5.3 Filtering and interaction

Maps become more useful when connected to slicers, charts, and tables. Selecting a product category can highlight affected locations; selecting a region can filter a trend chart. Cross-filtering must be tested because the apparent spatial pattern can change with every filter.

Include a visible indication of filter state. A map that shows only one quarter or one customer segment without clearly communicating that restriction can be misinterpreted as a complete view.

6. A Practical Construction Method

The following method is suitable for a certificate-level Power BI project.

Step 1: Define the decision question

Begin with a business question, such as: “Which service areas have high demand but insufficient branch capacity?” Avoid beginning with the intention to “make a map.”

Step 2: Identify the spatial unit

Decide whether the unit is a customer, branch, postal code, municipality, route, or territory. The spatial unit must match the decision.

Step 3: Select the measure and denominator

Choose total, average, rate, or index deliberately. If populations differ, consider a normalized measure. Document the source and date of any denominator.

Step 4: Validate locations

Check categories, coordinates, missing values, duplicate sites, and unusual points. Compare a sample with authoritative geographic references.

Step 5: Build the simplest useful visual

Start with points or polygons and one primary measure. Add complexity only when it answers a real question.

Step 6: Add context

Use boundaries, labels, reference layers, tooltips, and comparison charts. Do not add layers merely because they are available.

Step 7: Test interpretation

Ask whether a user can identify the measure, time period, geography, and scale. Test small screens, filtered states, and accessibility features.

Step 8: Document assumptions

State the geocoding source, geographic definitions, time period, missing-data treatment, and rate denominator.

7. Worked Example: Regional Service Demand

A service company has 12 branches. Management wants to identify regions where demand is high relative to branch capacity. The dataset includes region, population, service requests, branch count, and coordinates for each branch.

Suppose Region A has 18,000 service requests and a population of 600,000. Its request rate per 100,000 residents is:

Suppose Region B has 12,000 requests and a population of 200,000:

Region A has more requests in absolute terms, but Region B has twice the rate. A map of totals may prioritize Region A for workload planning, while a rate map may identify Region B as having greater relative demand. Neither is universally correct; each answers a different question.

A suitable report could contain:

• An Azure Maps point layer showing branch locations.
• Bubble size representing branch capacity.
• A region-level choropleth showing requests per 100,000 residents.
• Tooltips showing total requests, rate, branch count, and average response time.
• A column chart comparing regions by capacity gap.

A capacity gap might be defined operationally as demand minus available capacity, but the definition must be documented. If capacity is measured in monthly appointments and demand is measured in requests, the units must be made comparable before subtraction.

8. Illustrative Case: Retail Network Expansion

Background

This illustrative case concerns “Northstar Retail,” a fictional retailer operating stores across three provinces. The company is considering two new locations. Executives initially requested a map of sales by province.

Problem

The first map used total revenue with a filled map. One large province appeared to be the best opportunity because it occupied substantial visual area and had the highest total sales. However, the province also had many more stores and a larger population. The map did not distinguish existing network scale from unmet demand.

Analysis

The analyst created the following measures:

• Total revenue by province
• Revenue per store
• Revenue per 100,000 residents
• Online orders without a nearby store
• Average delivery time

Coordinates were used for existing stores. A province boundary layer displayed normalized demand, while a point layer showed stores. Tooltips combined revenue, store count, population, online orders, and delivery time.

The analysis found that the largest province had high total sales but moderate revenue per store. A smaller province had fewer total sales but a high rate of online orders, long delivery times, and no stores in two population centres.

Approach and solution

The final report used:

1. A choropleth of online orders per 100,000 residents.
2. A point layer for stores, sized by annual revenue.
3. A reference layer for population centres.
4. A table ranking candidate cities by demand, distance to the nearest store, and estimated operating cost.
5. A note explaining that the map identified candidates for further feasibility analysis rather than proving that a new store would be profitable.

Outcome

Management selected one candidate in the smaller province for detailed assessment and postponed expansion in the large province. The decision was not based on colour intensity alone; it combined spatial demand, network coverage, logistics, and financial analysis.

Learning points

• Totals and rates answer different questions.
• A map should display the spatial unit relevant to the decision.
• Point and polygon layers can complement one another.
• Geographic evidence supports a decision but does not replace financial or operational evaluation.

9. Responsible Spatial Comparison

Maps influence attention and can create strong impressions. Responsible practice requires the analyst to distinguish measurement from visual effect.

9.1 Avoid comparing incompatible areas

A large region is not necessarily more important than a small region. Area, population, store count, and exposure should be considered. If the aim is to compare customer intensity, use a suitable rate rather than raw totals.

9.2 Avoid misleading colour scales

A diverging colour scale is appropriate when there is a meaningful midpoint, such as a target or zero change. A sequential scale is appropriate for values progressing from low to high. Colour classes should not imply a level of precision unsupported by the data.

Red–green combinations can create accessibility problems for users with colour-vision deficiency. Use contrast, labels, patterns where supported, or supplementary tables.

9.3 Recognize the modifiable areal unit problem

Results can change when data is aggregated into different boundaries. A pattern visible by province may disappear by municipality, or vice versa. This is a general limitation of area-based analysis. Analysts should state the geographic level and avoid assuming that an aggregate pattern applies to every individual within the area.

9.4 Handle uncertainty and missing data

Missing locations should not silently disappear. Report the number or percentage of records that could not be mapped. Geocoded points may also have varying accuracy. A city-centre approximation is not equivalent to a verified site coordinate.

9.5 Protect privacy

Maps can expose customer residences, patient locations, employee movements, or incident sites. Use aggregation, suppression, masking, spatial jittering, or minimum-count rules where appropriate. Exact coordinates should not be published merely because they are available. Access controls and row-level security may be needed for sensitive reports.

9.6 Governance and compliance

Organizations should document data ownership, geocoding sources, licensing conditions, refresh dates, retention rules, and permitted audiences. Location data may be personal data under applicable privacy law when it can identify or reasonably single out a person. Legal and organizational requirements vary by jurisdiction; analysts should consult their privacy and governance policies.

10. Accessibility and Communication

A map should not be the only way to obtain the conclusion. Provide an accompanying table or textual summary, meaningful titles, alt text where supported, keyboard-accessible controls, and sufficient contrast. Tooltips should explain abbreviations and units.

A good title includes the measure, geography, and period—for example, “Service requests per 100,000 residents by district, January–June 2025.” A weak title such as “Regional Performance” leaves too much interpretation to the reader.

11. Limitations and Contemporary Developments

Spatial visualization has several limitations:

• Geocoding may be incorrect or incomplete.
• Boundary definitions may change over time.
• Aggregation can hide local variation.
• Visual proximity does not prove causal connection.
• A route line may be schematic rather than an actual travelled path.
• Basemap availability and service features may depend on organizational settings.
• Large datasets may reduce performance and create visual overplotting.

Contemporary practice increasingly combines geographic visuals with spatial statistics, real-time location streams, mobility data, and AI-assisted insights. Such tools can help detect clusters or anomalies, but automated suggestions require validation. AI may identify an apparent spatial pattern without understanding data quality, boundary effects, or privacy implications. Human review remains necessary.

Best-Practice Checklist

Before publishing a geographic report, confirm that:

• The location fields have correct data categories.
• Latitude and longitude are not reversed.
• Ambiguous place names include geographic context.
• The visual uses the appropriate spatial unit.
• Totals, averages, rates, and denominators are clearly distinguished.
• The legend and colour scale are interpretable.
• Filters and time periods are visible.
• Missing and low-confidence locations are documented.
• Sensitive locations are protected.
• A table or text alternative supports the map.
• The map answers a defined decision question.

Chapter Summary

Geographic visualization connects measures with place, enabling analysts to examine distribution, concentration, coverage, movement, and regional difference. Reliable mapping begins with accurate location fields, suitable data categories, validated coordinates, and meaningful geographic hierarchies. Power BI map visuals can represent points, polygons, density, boundaries, and routes; Azure Maps can combine analytical layers with geographic context.

The choice between totals, rates, averages, and normalized measures is central to responsible comparison. Choropleths are appropriate for area-level values, while point maps are appropriate for individual locations. Layers, drill-down, filters, and tooltips should clarify the decision rather than add decoration. Finally, maps must address uncertainty, accessibility, privacy, licensing, and the risk that visual patterns will be interpreted as causal evidence.

Key Terms

Geocoding: Converting a place description or address into geographic coordinates.

Latitude: Angular position north or south of the equator.

Longitude: Angular position east or west of a reference meridian.

Coordinate pair: A latitude and longitude used to identify a location.

Geographic hierarchy: Ordered levels of place, such as country, state, city, and site.

Point data: Spatial observations represented as individual locations.

Line data: Spatial features represented as paths or connections.

Polygon data: Spatial features represented as bounded areas.

Choropleth: A map that shades areas according to a measured value.

Heat map: A visual representation of concentration or intensity.

Reference layer: A contextual spatial layer showing boundaries, roads, or other features.

Data category: A semantic classification that tells Power BI how a field should be interpreted.

Drill-down: Moving from a broad geographic level to a more detailed level.

Denominator: The population, exposure, or base used to normalize a measure.

Rate: A measure calculated relative to a relevant denominator.

Overplotting: Visual concealment caused by many marks occupying the same area.

Modifiable areal unit problem: The possibility that results change when spatial boundaries or aggregation levels change.

Spatial jittering: Deliberately displacing a location slightly to reduce identification risk or overlap.

Basemap: Background geographic information displayed beneath analytical layers.

Review Questions

Short Answer Questions

1. What is the difference between point data and polygon data?
2. Why should latitude and longitude fields be validated before mapping?
3. What is a geographic hierarchy?
4. When is a rate more informative than a total?
5. Name two privacy risks associated with geographic visualization.

Long Answer Questions

1. Explain the process of preparing location data for a Power BI map, from data categorization through validation.
2. Compare point maps, choropleths, heat maps, and shape maps, including an appropriate use for each.
3. Explain how Azure Maps layers can be combined to support operational analysis.
4. Discuss why totals may produce misleading regional comparisons and describe suitable alternatives.
5. Describe the role of accessibility, governance, and privacy in publishing spatial reports.

Analytical / Application Questions

1. A city has 40,000 incidents and a population of 2,000,000. A neighbouring city has 12,000 incidents and a population of 300,000. Calculate the incident rate per 100,000 residents for each city and explain which city has the higher relative incidence.
2. Design a Power BI report for a delivery company that wants to identify areas with long delivery times and low depot coverage. Specify the spatial unit, measures, map layers, filters, and safeguards you would use.
3. A filled map shows the highest sales in the largest geographic region. Explain at least three reasons why this may not identify the best location for expansion, and propose a more reliable analytical approach.

Multiple Choice Questions with Answers

Question 1

Which field combination most directly reduces ambiguity when mapping a city name?

A. City and product category

B. City and country

C. City and sales amount

D. City and order status

Correct Answer: B. City and country

Question 2

Which visual is generally most suitable for comparing a rate across administrative areas?

A. Choropleth map

B. Route map

C. Point map with no measure

D. Timeline chart

Correct Answer: A. Choropleth map

Question 3

What does a latitude value primarily indicate?

A. East–west position

B. Population density

C. North–south position

D. Map zoom level

Correct Answer: C. North–south position

Question 4

A region has 18,000 requests and a population of 600,000. What is its request rate per 100,000 residents?

A. 300

B. 1,800

C. 3,000

D. 30,000

Correct Answer: C. 3,000

Question 5

Which practice best protects privacy in a map of sensitive customer locations?

A. Publish every exact coordinate

B. Remove all geographic context

C. Use aggregation or suppression rules

D. Increase map zoom to maximum

Correct Answer: C. Use aggregation or suppression rules

Question 6

What is the main purpose of a reference layer in Azure Maps?

A. To replace all measures

B. To provide geographic context such as boundaries or roads

C. To calculate financial totals

D. To correct missing sales records

Correct Answer: B. To provide geographic context such as boundaries or roads

Question 7

Why can a map of total sales favour a large province misleadingly?

A. Large provinces cannot contain stores

B. Totals may reflect population and store count rather than opportunity

C. Totals are always rates

D. Province boundaries automatically remove data

Correct Answer: B. Totals may reflect population and store count rather than opportunity

Question 8

Which measure is most appropriate for comparing incident burden across populations of different sizes?

A. Incident count only

B. Incident rate using a relevant population denominator

C. Map area in square kilometres only

D. Number of colours in the legend

Correct Answer: B. Incident rate using a relevant population denominator

Question 9

What is overplotting?

A. A missing geographic category

B. Excessive map licensing

C. Visual concealment caused by overlapping marks

D. A geographic hierarchy with too few levels

Correct Answer: C. Visual concealment caused by overlapping marks

Question 10

Which statement about a spatial pattern is most responsible?

A. A visible pattern automatically proves causation

B. A map can identify a pattern, but other analysis is needed to establish causes

C. A larger symbol always represents a larger area

D. All mapped coordinates have equal accuracy

Correct Answer: B. A map can identify a pattern, but other analysis is needed to establish causes