Data visualisation is the representation of data in graphical or visual form to make patterns, comparisons, relationships, distributions and changes easier to understand. Common forms include bar charts, line charts, scatter plots, histograms, box plots, heat maps, maps, area charts and carefully designed dashboards. Visualisation does not merely decorate numerical information; it provides a perceptual interface through which people can inspect data. A well-designed visualisation can reveal trends or anomalies that are difficult to identify in a table, while a poorly designed one can obscure important information or create misleading impressions.
Different chart types are suited to different analytical questions. Bar charts are commonly useful for comparing discrete categories, line charts for displaying changes over an ordered sequence such as time, scatter plots for examining relationships between quantitative variables, histograms for examining distributions and box plots for summarising distributions and identifying potential outliers. Maps can be appropriate when geographic location is central to the question, while network diagrams can represent relationships among connected entities. Choosing a visual form should begin with the analytical task and characteristics of the data rather than with the visual appearance of a particular chart type.
Effective data visualisation requires appropriate scales, axes, labels, units and context. A truncated axis can exaggerate apparent differences in some chart types, while an inappropriate scale can make meaningful variation difficult to see. Percentages should normally identify their denominator, and time-series charts should make the time interval clear. Legends and annotations should explain information that cannot be understood from the visual structure alone. Titles should communicate what is being shown rather than merely naming the dataset. Accurate visualisation preserves the relationship between the displayed representation and the underlying numerical values.
Visual encoding determines how data values are represented through position, length, area, angle, colour, shape or other graphical properties. Human perception generally makes some encodings easier to compare accurately than others. For example, positions along a common scale are typically easier to compare precisely than areas or angles. This is one reason bar charts can often support clearer quantitative comparisons than pie charts when there are many categories. Colour can distinguish categories or represent magnitude, but it should be used carefully because colour perception varies and some colour combinations are difficult for people with colour-vision deficiencies to distinguish.
Dashboards combine multiple visualisations and often provide interactive filtering, drill-down and monitoring features. They are commonly used to display key performance indicators, operational metrics, trends and exceptions. A dashboard should have a clear purpose and audience because displaying too many unrelated charts can increase cognitive load rather than improve understanding. Operational dashboards may prioritise rapid detection of abnormal conditions, whereas analytical dashboards may provide richer exploration. Good dashboard design also considers refresh frequency, data freshness, definitions and what actions users are expected to take after identifying a pattern.
Data visualisation is closely connected to exploratory data analysis and communication of analytical results. During exploration, analysts may create many temporary charts to understand distributions, relationships and anomalies. For communication, the final visualisation may be simplified to emphasise the findings relevant to a particular audience. A visualisation intended for a technical analyst can expose detailed scales and statistical information, while one intended for senior decision-makers may focus on a small number of clearly defined indicators. In both cases, simplification should remove unnecessary complexity rather than remove information that changes the interpretation.
Visualisations can mislead without containing technically false numbers. Examples include inappropriate axis ranges, selective time periods, inconsistent category definitions, three-dimensional effects that distort perceived size, excessive decorative elements and colour choices that imply unsupported categories. A chart can also encourage a causal interpretation of a simple correlation. Visualisations should therefore be accompanied by sufficient context and, where relevant, methodological information. Users should be able to determine what population, period, units and definitions the displayed values represent.
Data visualisation has evolved from static statistical graphics and printed charts into interactive, web-based and real-time systems. Modern tools allow users to filter datasets, inspect individual records, combine multiple dimensions and display continuously updated information. Visualisation is used in science, journalism, business intelligence, engineering, finance, public administration, education and many other fields. Its fundamental purpose remains the same: represent quantitative or categorical information in a form that supports human perception and reasoning while preserving the important characteristics and limitations of the underlying data.
