7. Visualizing Trends and Patterns

A practical guide to choosing, designing, evaluating, and making accessible visualizations that communicate patterns and trends accurately.

Start with the question

A visualization should answer a clear question rather than display every available field. Begin by identifying the main task:

Comparison: Which category is largest or smallest?
Change over time: Is a value increasing, decreasing, or remaining stable?
Distribution: How are values spread out?
Relationship: Do two variables appear to change together?
Composition: How does a total divide into parts?
Anomaly: Which value is unusually high or low?

Then select the variables and comparisons that reveal the finding. Derived values, such as percentage change, rates per person, or differences from an average, may communicate a pattern more clearly than raw values. For example, a list of monthly library visits stores the data, but a makes rises and falls easier to detect. If the question is instead which grade uses the library most, a category comparison calls for a different visual form.

Takeaway: Define the question first; choose the visual second.

Use tables for precision

A is the most useful format when exact values, detailed lookup, or comparisons across several variables matter. Give the a descriptive title and label every row and column clearly. Include units, use consistent rounding, and explain abbreviations, missing values, and unusual symbols. Remove columns that do not help answer the question.

Tables are also an important text-based alternative to charts. Essential information should remain available in a non-visual format so that people using screen readers, small screens, or other access technologies can use the data. A chart can show the broad pattern while a provides precise supporting values.

Takeaway: Use tables for precision, lookup, and accessible access to the underlying values.

Match the visual to the evidence

Match the chart to the structure of the data and the question being asked.

A or column chart compares distinct categories. Ordering bars from highest to lowest can make a ranking easier to see.
A shows values across an ordered variable, usually time. It is suited to trends, cycles, sudden changes, and stable periods.
A shows paired values for two numerical variables. It can reveal association, clusters, and outliers, but association alone does not prove causation.
A groups numerical values into intervals to show concentration, skew, gaps, or multiple clusters.
A box plot summarizes a distribution with the median, quartiles, and possible outliers, making it useful for comparing the spread of groups.
Stacked bars show how categories contribute to a total, but interior segments that do not share a baseline can be difficult to compare precisely.
Maps are appropriate when location is central to the question. Geographic information should also be available as text or a .

The choice of bin width, scale, summary measure, or chart type can change what viewers notice. Use a simpler chart or a when precise comparisons between parts are important.

Takeaway: Let the data structure and analytical question determine the visual form.

Build focused dashboards

A combines multiple visual elements to provide an overview of a topic. It is most useful when data are updated regularly and users need to monitor a focused set of important indicators.

Effective dashboards:

Put the most important information in the most prominent position.
State the purpose with a clear title.
Show dates, units, definitions, and relevant context.
Group related information together.
Use filters only when they answer a real user need.
Make trends and exceptions easy to identify.
Provide accessible tables or text alternatives.
Are tested with representative users and reviewed when the data or purpose changes.

Too many charts can overload the reader, while interactive controls can hide important information behind clicks. A focused set of charts with explanatory text may be clearer than one crowded screen.

Takeaway: Interactivity should support understanding, not hide essential evidence.

Design for clarity

Design choices should direct attention to the intended comparison without changing the underlying data.

Titles and annotations: Use a descriptive title that communicates what viewers should notice. Add subtitles for the population, period, or measurement method, and annotations for events that may explain unusual changes.
Axes and scales: Label variables and units, order time correctly, use readable and consistently spaced tick marks, and use comparable scales when comparing panels.
Ordering and grouping: Sort categories by size to emphasize ranking, use chronological order to emphasize change, or group related categories logically.
Color and emphasis: Use color purposefully and consistently. Do not rely on color alone; add labels, patterns, symbols, or direct annotations when needed.
Decoration: Avoid three-dimensional effects, heavy borders, shadows, textures, and dark gridlines that compete with the data.

A clear title such as “Library visits increased during the first three months” communicates more than a generic title such as “Library Visits.”

Takeaway: Labels, ordering, scale, and restrained emphasis help readers notice the intended finding.

Check for misleading impressions

A visualization can mislead even when every plotted value is technically correct. Watch for these problems:

: A small difference can appear dramatic when a does not use a common zero baseline.
Unequal intervals: Irregular spacing on an axis can suggest changes that are not present in the data.
Inappropriate chart types: A map or pie-style display may be unsuitable when exact comparisons are needed.
Dual axes: Two scales can make differently scaled or unrelated trends appear connected.
Inconsistent scales: Similar charts with different axes can make comparisons unfair.
Selective time periods: Unusual starting or ending dates can hide part of a trend.
Overloaded design: Too many categories, colors, or labels can obscure the message.
Unexplained averages: A mean can hide variation, outliers, or differences between groups.
Missing context: Raw counts may be unfair to compare when group sizes differ; rates or percentages may be more appropriate.
Unclear uncertainty: A precise-looking estimate can conceal sampling error or other limitations.

For example, two neighborhoods may report the same number of bicycle thefts, but the neighborhood with a larger population may have a lower theft rate. A fair comparison may require a population-adjusted rate rather than raw counts.

Takeaway: Inspect scales, context, denominators, time periods, and uncertainty before accepting the visual message.

Follow a practical visualization process

Use this process to turn data into visual evidence:

Define the question. State the comparison, trend, relationship, composition, distribution, or anomaly to investigate.
Check the data. Look for missing values, duplicates, inconsistent units, and unusual records.
Choose the visual form. Match a chart or to the data and the question.
Select the comparison. Decide whether to compare categories, time periods, regions, totals, rates, or averages.
Design for clarity. Use readable labels, suitable scales, meaningful colors, and limited decoration.
Add context. Include units, dates, definitions, and notes about uncertainty or missing data.
Provide an alternative format. Include a or text description for essential information.
Test the result. Ask someone unfamiliar with the data what conclusion they draw, then revise the visual if the intended pattern is not clear.

The best visualization is not the most elaborate one. It is the one that communicates relevant evidence accurately, clearly, and accessibly.

Final takeaway: A successful visualization aligns the question, data, visual form, design, context, and audience.