Data Visualization Techniques | Histograms, Line Charts, Scatter Plots, and Applications

What Is Data Visualization? Definition and Core Concept

Data visualization is the graphical representation of numerical and categorical information so that patterns, trends, and outliers become visible at a glance instead of buried inside a spreadsheet. It converts columns of numbers into shapes, positions, and colors that the human visual system can process almost instantly. A well-built chart does in half a second what a table of a thousand rows cannot do at all: it lets a reader see the shape of the data.

The discipline sits at the intersection of statistics, design, and cognitive science. As Principles of Effective Data Visualization explains, the basic scatterplot remains highly effective on its own, and layering extra information through point size, symbol, and color can highlight additional patterns without cluttering the underlying data. That balance between showing enough and showing too much is the central tension every visualization technique has to resolve.

Every chart is built from the same raw materials: data, encoding, and perception. Data is the underlying numbers, whether they come from an experiment, a survey, or a company database. Encoding is the mapping of those numbers onto visual properties like position, length, angle, or color. Perception is what happens in the viewer’s brain once the encoding reaches their eyes. Choosing the right chart type means choosing an encoding that the human visual system can decode accurately and quickly. If you are still building the underlying dataset, finding a reliable dataset is the first practical step before any chart gets drawn.

What Makes a Data Visualization “Good”?

A good chart is judged by whether it lets the intended audience answer a specific question accurately and quickly, not by how visually elaborate it looks. A dense 3D exploded pie chart may look impressive in a slide deck, but if a reader cannot rank the slices correctly, the chart has failed at its one job. Effectiveness, not decoration, is the real measure of quality.

This is why data visualization is taught alongside descriptive statistics in most introductory statistics courses. A histogram and a mean are answering related but different questions — the mean gives you one number, the histogram shows you the entire shape the number was drawn from. Students who understand measures of variability alongside chart construction build a much deeper intuition for what the data is actually doing.

Exploratory vs Explanatory Visualization: Two Different Jobs

Data visualization splits into two distinct modes with different goals. Exploratory visualization is what an analyst does privately while investigating a new dataset — fast, rough charts meant to surface patterns, check assumptions, and catch data-quality problems. Explanatory visualization is the polished, audience-facing chart built afterward to communicate one specific finding clearly. Confusing the two is one of the most common errors in student assignments and business reports alike: a chart built for personal exploration, dense with every variable at once, rarely communicates well to an outside reader.

Why Data Visualization Matters: The Science of Visual Perception

The reason some charts work better than others is not a matter of taste. It is rooted in decades of research on how accurately humans judge different visual properties. Statisticians William Cleveland and Robert McGill ran the foundational experiments in 1984, ranking visual encodings by how precisely people could compare them. Position along a common scale ranked highest. Length ranked next. Angle, area, and color saturation ranked considerably lower, and volume ranked lowest of all.

This ranking explains almost every rule of thumb in chart design. Bar charts outperform pie charts because length judgments beat angle judgments. Scatter plots outperform bubble charts for precise comparison because position beats area. As the research on graphical perception summarizes, length is more effective than area and volume for accurate comparison, which is exactly why so much of the visualization literature keeps returning to line charts and bar charts for precise numeric reading.

What Does “Perceptually Accurate” Actually Mean?

A perceptually accurate chart is one where the visual difference a reader perceives matches the actual numeric difference in the underlying data. If one bar is twice as tall as another, the underlying value really is twice as large, and the reader’s eye estimates that ratio correctly. Pie charts break this rule more easily than bar charts because judging two similar angles accurately is a much harder visual task than judging two lengths.

Recent research complicates the simple story somewhat. One study on pie chart decision-making found no significant difference between pie charts and bar charts on a real-world academic ranking task with over 300 participants, challenging the assumption that pie charts always hurt high-stakes decisions. The nuance matters for students: perceptual rankings are strong general guidance, not an absolute law that applies identically to every task and every audience.

Color, Accessibility, and Perceptually Uniform Palettes

Color is one of the most misused encodings in student and business charts. The default rainbow or “jet” colormap creates false visual boundaries because it is not perceptually uniform — equal numeric steps do not look like equal visual steps. Modern practice favors perceptually uniform, colorblind-safe palettes such as Viridis, Plasma, and Cividis, along with the ColorBrewer palettes originally designed for cartography and now standard across scientific plotting libraries.

Roughly 8% of men of Northern European descent have some form of color vision deficiency, which means any chart relying purely on red-green contrast risks losing a meaningful share of its audience. Pairing color with a second encoding, such as line style or direct labeling, protects the chart’s message regardless of how a given viewer perceives color.

Histograms: Visualizing Distribution and Frequency

A histogram is a chart that groups a single continuous variable into consecutive intervals, called bins, and draws a bar for each bin whose height represents how many observations fall inside it. Unlike a bar chart, the bars in a histogram touch, because the underlying variable is continuous and the bins represent an unbroken range rather than separate categories. A histogram answers one question extremely well: what does the shape of this variable’s distribution look like?

That shape question matters enormously before running almost any statistical test. Checking whether data looks roughly bell-shaped, skewed, or bimodal tells you whether assumptions behind a t-test or a parametric model are reasonable, or whether you need a non-parametric test instead. As one applied guide to scientific charting puts it, visualizing a distribution with a histogram before running any statistical test reveals whether the data is normally distributed, bimodal, heavily skewed, or contains outliers, all of which affect which tests remain valid.

How Bin Width Changes the Story

The single most important decision in building a histogram is choosing the bin width. Too few, wide bins smooth away real structure, such as a second peak in the data. Too many, narrow bins introduce noisy, jagged bars that make random sampling variation look like meaningful pattern. Common rules of thumb, such as Sturges’ rule or the Freedman-Diaconis rule, provide sensible starting points, but the right choice ultimately depends on sample size and the question being asked. Students working through mean, median, and mode calculations in Excel often build a histogram in the same worksheet to sanity-check whether the mean is even a meaningful summary of the data.

Reading Skewness and Kurtosis From the Shape

Histograms make skewness and kurtosis visible in a way that a single summary statistic cannot. A right-skewed histogram has a long tail stretching toward higher values, common in variables like household income or hospital wait times. A left-skewed histogram has the opposite tail. A histogram with heavier tails than a normal bell curve signals high kurtosis, meaning extreme values are more common than a normal distribution would predict. For a deeper treatment of exactly how these shapes are quantified, see the guide to normal distribution, kurtosis, and skewness.

Histogram vs Density Plot vs Box Plot

A density plot smooths the histogram into a continuous curve using kernel density estimation, trading some transparency about raw bin counts for a cleaner visual read of the overall shape. A box plot compresses the same distribution into five summary numbers: minimum, first quartile, median, third quartile, and maximum, plus flagged outliers. Box plots are excellent for comparing several groups side by side but hide multimodal shapes that a histogram would reveal instantly. Recent ecological research on replacing bar charts with histograms found that combining a histogram or density plot with a univariate scatter of the raw points gives readers the fullest possible picture of a small dataset’s true shape.

Line Charts: Visualizing Trends Over Time

A line chart plots a continuous variable against an ordered axis, almost always time, and connects consecutive points with straight or smoothed segments. The connecting line itself is a visual claim about continuity between measurements, which is why line charts belong to ordered, typically time-based, data and not to unordered categories. The line graph was invented by the Scottish engineer and economist William Playfair in 1786, making it one of the oldest deliberate statistical graphics still in everyday use, according to the graphical perception literature that traces the technique’s origins.

Line charts are the standard way to present time series analysis results, whether the underlying series is stock prices, monthly enrollment figures, or temperature readings. Because position along a shared axis is the most accurately perceived visual encoding, a well-drawn line chart lets readers read off both the overall trend and the fine-grained fluctuations in a single glance.

Single Series vs Multiple Series: When Lines Compete

A single line chart is nearly always unambiguous. The moment a second or third line is added, readers must track and compare multiple trajectories simultaneously, and research on graphical perception for time series shows that different visual designs support different tasks: position-based line charts are strong for finding an extremum or a specific value, while color-field or heatmap-style alternatives can outperform for spotting an overall pattern across many series at once. Beyond four or five lines, a standard line chart usually becomes what practitioners call “spaghetti,” and small multiples, where each series gets its own mini panel, communicate more clearly than cramming everything onto one axis.

Smoothing, Downsampling, and the Risk of Distortion

Real-world time series are often noisy, and analysts frequently apply a moving average or other smoothing technique to reveal the underlying trend. This is a legitimate and common practice, but it comes with a real interpretive cost. Research on smoothing and time series interpretation, using COVID-19 case-count data as a case study, found that smoothing choices meaningfully change how readers interpret the trajectory of an outbreak, sometimes making a plateau look like a decline or vice versa. Whenever a chart applies smoothing, transparency about the smoothing window and method is essential for honest reporting, a principle covered in more depth in the guide to transparent results reporting.

A related issue arises with very large datasets. When a time series contains millions of points, every plotting library has to downsample before rendering, and the algorithm chosen for that downsampling can visually distort the shape of spikes and drops. Work on data point selection for line charts found that naive downsampling methods can miss short-lived peaks entirely, while purpose-built algorithms preserve visual fidelity at scale. For coursework involving smaller datasets, this is less of a concern, but it becomes critical the moment a project scales to production-level data volumes typical of a machine learning pipeline.

The Y-Axis Baseline Rule

Scatter Plots: Visualizing Relationships and Correlation

A scatter plot places each observation as a point on two axes representing two continuous variables, revealing the direction, strength, and shape of the relationship between them. Unlike a line chart, the axis does not need to represent time or any inherent ordering — a scatter plot works equally well for height versus weight, advertising spend versus sales, or study hours versus exam score.

Scatter plots are the visual foundation of simple linear regression. Before fitting a line through any dataset, plotting the raw points first is essential, because a regression line can be fit through data with almost any underlying shape, including data where a straight line is completely the wrong model. As Doing Better Data Visualization argues, the scatterplot is superior to a line plot for showing a relationship because it displays both the trend and the individual observations driving that trend, and adding marginal histograms along each axis further improves the chart by showing the distribution of each variable alone.

Reading Correlation Direction and Strength From the Cloud

A scatter plot where points trend from lower-left to upper-right indicates a positive relationship: as one variable rises, so does the other. A trend from upper-left to lower-right indicates a negative relationship. The tightness of the cloud around an imaginary line indicates strength: a thin, cigar-shaped cloud signals a strong relationship, while a wide, circular cloud signals a weak or nonexistent one. These visual impressions correspond directly to the numeric measures covered in the guide to covariance and correlation.

Handling Overplotting in Large Datasets

When a scatter plot contains thousands or millions of points, individual markers overlap so heavily that the chart becomes a solid blob, hiding the true density pattern. Analysts address this with several techniques: reducing point opacity so darker regions signal higher density, switching to a hexbin or 2D histogram that bins the plane into small hexagonal or square cells, or using a contour plot that turns the point cloud into a topographic-style density map. Research on interactive scatterplot evaluation found that letting users filter and interact with a multiple-scatterplot layout increased both task confidence and stated preference compared to a single static view, which is why modern dashboards frequently make scatter plots interactive rather than static.

Bubble Charts: The Scatter Plot’s Third Dimension

A bubble chart extends the scatter plot by mapping a third continuous variable to the size of each point, and sometimes a fourth to color. This is powerful for showing relationships like GDP per capita versus life expectancy, sized by population, but because area is a less accurately perceived encoding than position, bubble charts should be reserved for cases where the third variable is genuinely secondary to the main x-y relationship, not the primary message of the chart.

Other Essential Chart Types: Bar Charts, Box Plots, and Heatmaps

Histograms, line charts, and scatter plots handle most statistical questions, but a complete visualization toolkit needs a few more specialized techniques for categorical comparisons, group distributions, and dense matrix data.

Bar Charts: The Workhorse of Categorical Comparison

A bar chart compares distinct, unordered or ordered categories using bar length, the second most accurately perceived visual encoding after position. Unlike a histogram, the bars in a standard bar chart do not touch, because the categories are discrete rather than continuous bins of a single variable. Bar charts are ideal for comparing sales across regions, survey responses across age groups, or scores across different treatment conditions.

Box Plots: Comparing Distributions Across Many Groups

A box plot, or box-and-whisker plot, is the standard scientific chart for comparing the distribution of a continuous variable across several categories at once. It displays the median as a center line, the interquartile range as a box, and the full spread with whiskers, flagging points beyond a defined range as outliers. As one guide to scientific charting notes, many statistics journals now explicitly prefer box plots over simple bar charts with error bars for continuous outcome data, because a box plot conveys median, spread, and outliers simultaneously, information a bar-and-error-bar chart cannot fully represent.

Heatmaps: Visualizing Dense Matrix Data

A heatmap encodes a matrix of numbers as color intensity across a grid, making it possible to scan hundreds or thousands of values at once. A correlation heatmap, where each cell represents a Pearson correlation coefficient colored on a diverging red-to-blue scale, is one of the most common uses in statistics and data science, letting an analyst instantly spot the strongest positive and negative relationships across an entire dataset before running any formal regression analysis.

Pair Plots and Scatterplot Matrices

A scatterplot matrix, or pair plot, arranges every pairwise combination of variables in a dataset into a grid, with histograms or density curves along the diagonal and scatter plots off the diagonal. This gives an analyst a full picture of every variable’s distribution and every pairwise relationship in a single figure, making it one of the fastest ways to begin exploratory analysis before dimensionality reduction techniques such as principal component analysis.

Pie Charts and Donut Charts: Use With Caution

Pie charts encode values as angle and area, both perceptually weaker than the length encoding used in bar charts. Data journalist and author Cole Nussbaumer Knaflic’s widely cited critique of pie charts, referenced across the visualization literature as “Death to Pie Charts,” argues that a simple bar chart nearly always communicates the same information more precisely. That said, more recent controlled research, including the study on real-world pie chart decisions, found no measurable harm from pie charts in a specific high-level ranking task, so the safest academic position is that pie charts are usable for two or three slices with clearly labeled values, but a bar chart is the more defensible default whenever precision matters.

How to Choose the Right Chart for Your Data

Chart selection is not a matter of preference. It follows directly from two questions: what type of data do you have, and what question are you trying to answer? Getting this pairing right is the single highest-leverage decision in the entire visualization process, more important than color choice, font, or software.

Start With the Data Type

The first fork in the decision tree is the underlying data type. Understanding the distinction between qualitative and quantitative data, and more specifically the four measurement scales covered in nominal, ordinal, interval, and ratio data, determines which chart family is even valid. A single continuous variable calls for a histogram or density plot. Two continuous variables call for a scatter plot. A continuous variable across time calls for a line chart. A continuous variable across discrete categories calls for a bar chart or box plot. Two categorical variables call for a grouped bar chart, mosaic plot, or heatmap.

Then Ask What Question the Chart Must Answer

Once the data type narrows the options, the specific analytical question finishes the decision. “How is this variable distributed?” points to a histogram. “Is this variable trending up or down?” points to a line chart. “Are these two variables related?” points to a scatter plot. “Which category performs best?” points to a bar chart. “How does this group’s spread compare to that group’s?” points to a box plot.

Matching Chart Choice to Statistical Method

Chart selection also connects directly to which statistical test or model is appropriate downstream. A scatter plot that reveals a clear linear trend supports moving forward with multiple linear regression, while a scatter plot showing a curved relationship signals that a polynomial or non-linear model is more appropriate. A histogram showing strong skew signals that a hypothesis test assuming normality may need a transformation or a non-parametric alternative first. Students unsure which test fits their data shape can walk through the framework in choosing the right statistical test.

Key Tools, Organizations, and Pioneers Shaping the Field

Data visualization has a specific intellectual history and a concrete ecosystem of software tools. Understanding both gives your academic writing more depth and helps you choose the right tool for a given assignment or project.

William Playfair (1759-1823): The Inventor of Statistical Graphics

William Playfair, a Scottish engineer and political economist, is widely credited with inventing the line graph, the bar chart, and the pie chart between 1786 and 1801. His 1786 Commercial and Political Atlas introduced the line graph specifically to make trade balance data over time understandable to readers without a background in mathematics. Nearly every chart type discussed in this article traces its lineage back to Playfair’s original insight that a visual encoding of numbers could communicate faster than a table.

John Tukey (1915-2000): Exploratory Data Analysis

John Tukey, the American mathematician and statistician at Princeton and Bell Labs, invented the box plot in the 1970s and coined the term “exploratory data analysis” to describe the practice of visualizing data before formal modeling. His 1977 book of the same name argued that graphical methods deserved equal footing with formal statistical inference, a philosophy that directly underlies the modern distinction between exploratory and explanatory visualization covered earlier in this guide.

Edward Tufte: The Data-Ink Ratio and Chartjunk

Edward Tufte, professor emeritus at Yale University, formalized the design principles that guide modern chart construction in his influential 1983 book The Visual Display of Quantitative Information. Tufte introduced the concept of the data-ink ratio, the proportion of a chart’s ink or pixels that directly represents data versus decoration, and coined the term “chartjunk” for unnecessary visual elements like heavy gridlines, 3D effects, and background textures that add no informational value. Maximizing the data-ink ratio while preserving clarity remains one of the most frequently cited design heuristics in the field.

Cleveland and McGill: The Science of Graphical Perception

William Cleveland and Robert McGill, statisticians at Bell Labs, published the 1984 study that experimentally ranked visual encodings by accuracy of human perception, discussed in detail earlier in this article. Their work moved chart design from aesthetic opinion to an empirically testable science, and it remains the theoretical backbone behind why position and length-based charts are generally recommended over area and angle-based ones.

Modern Visualization Software: Tableau, Power BI, and Programming Libraries

Tableau and Microsoft Power BI are the two dominant business intelligence platforms, both built around drag-and-drop chart construction for non-programmers, widely used across finance, marketing, and operations teams in the United States and United Kingdom. For statistical and academic work, programming libraries offer far greater flexibility: Matplotlib and Seaborn in Python, ggplot2 in R, and D3.js for custom interactive web visualizations are the standard toolkits taught in university statistics and data science programs. Choosing between a drag-and-drop tool and a coding library often comes down to whether the underlying analysis needs the reproducibility and version control that scripted code naturally provides.

Springer, Nature, and the Publication Standards Movement

Academic publishers increasingly enforce specific visualization standards for submitted research. Nature Cell Biology published a formal checklist condensing design, perception, and accessibility principles specifically to help researchers improve the clarity of published figures, reflecting a field-wide push toward more rigorous and honest data presentation standards across scientific journals.

Data Visualization in Statistics: Distributions, Outliers, and Regression

Data visualization is not a decorative add-on to statistics; it is a working diagnostic tool at nearly every stage of an analysis, from checking assumptions before a test to communicating a final result.

Checking Normality Before a Parametric Test

Many common statistical tests, including the t-test and standard regression models, assume the underlying data or residuals are approximately normally distributed. A histogram or a Q-Q plot, which plots sample quantiles against theoretical normal quantiles, is the fastest way to visually check this assumption before trusting a p-value. Understanding z-scores alongside these visual checks helps students connect the graphical pattern to the underlying probability theory.

Spotting Outliers Visually

Box plots and scatter plots are the two fastest ways to spot outliers before they distort a mean, a regression slope, or a correlation coefficient. A single extreme point in a scatter plot can single-handedly create the appearance of a relationship where none exists, or hide a real relationship among the remaining points. Formal outlier-sensitive diagnostics build on this visual first pass, covered in depth in the guide to residual analysis.

Visualizing Regression Assumptions

After fitting a regression line, a residual plot, which graphs the leftover error for each observation against the predicted value, is the standard way to check whether the model’s assumptions hold. A residual plot with no visible pattern supports the model; a residual plot with a curve, funnel shape, or trend signals a violated assumption, whether non-linearity, heteroscedasticity, or a missing variable. These checks are covered comprehensively in regression model assumptions and are essential before reporting any confidence interval around a regression coefficient.

Visualizing the Central Limit Theorem

One of the most common visualization exercises in an introductory statistics course is simulating repeated sample means from a non-normal population and plotting the resulting distribution as a histogram. The resulting bell shape, regardless of the original population’s shape, is the most direct visual proof of the central limit theorem, and this exercise is a staple of both AP Statistics and university-level courses because it turns an abstract theorem into something visibly true.

Choosing Between Descriptive Charts and Inferential Charts

A histogram or bar chart of raw sample data is a descriptive tool, summarizing what was actually observed. A chart showing a confidence interval, a p-value distribution, or a fitted regression line with a shaded uncertainty band is an inferential tool, making a claim about the broader population beyond the sample itself. Understanding this distinction, covered fully in descriptive vs inferential statistics, prevents a common student error of presenting a purely descriptive chart as if it proves a population-level claim.

Common Data Visualization Mistakes and Misleading Graphs

A chart can be technically accurate in its numbers while still misleading a reader through design choices. Recognizing these patterns protects both academic integrity and professional credibility.

The Truncated Y-Axis

Starting a bar or line chart’s y-axis at a value other than zero, without a clear visual break, exaggerates the apparent difference between values. A change from 98% to 99% looks dramatic on an axis running from 97 to 100, but trivial on an axis running from 0 to 100. This is the single most common way honest data becomes a misleading chart, whether in a student assignment, a corporate earnings deck, or a political advertisement.

Dual Y-Axes and False Correlation

Placing two different variables on two separate y-axes, scaled independently, allows a chart maker to make almost any two lines appear to move together, simply by adjusting each axis’s range. Because readers instinctively compare the visual position of the two lines rather than reading the actual axis values, dual-axis charts are one of the easiest ways to visually imply a correlation that does not exist in the underlying numbers.

Cherry-Picked Time Windows

A line chart’s story changes dramatically depending on the start and end date chosen. Showing only a short window that happens to trend favorably, while omitting the longer context, is a common way to mislead without technically fabricating any numbers. This overlaps closely with the broader problem of misuse of statistics through selective reporting, where a favorable-looking result is presented while less favorable framings of the same data are quietly left out.

Area and Volume Distortion in 3D Charts

3D bar and pie charts distort the true value of each element because of perspective. A bar further from the viewer appears shorter than an identical bar closer to the camera, even though the underlying data value is the same. Because area and volume are already among the least accurately perceived visual encodings, adding a third, purely decorative dimension compounds the distortion rather than adding useful information — a textbook case of Tufte’s “chartjunk.”

Ignoring Sample Size in the Visualization

A bar chart comparing group averages looks identical whether each group has five observations or five thousand, unless the chart explicitly shows sample size or uncertainty. Two groups with very different reliability can appear equally confident on a plain bar chart. Adding error bars, confidence intervals, or the raw data points as an overlay, and grounding the underlying sampling method in the chart’s caption, prevents readers from over-trusting a comparison built on a small or unevenly sized sample.

Data Visualization Across Industries: Real-World Applications

Data visualization techniques extend far beyond the statistics classroom into nearly every field of professional and academic work. The examples below show how histograms, line charts, and scatter plots translate into practical decisions across the U.S. and UK.

Business Intelligence and Financial Reporting

Corporate dashboards built in Tableau or Power BI rely heavily on line charts for revenue trends, bar charts for department comparisons, and heatmaps for identifying underperforming regions at a glance. Financial analysts use scatter plots to examine relationships between marketing spend and customer acquisition, directly applying the same regression analysis concepts taught in introductory statistics courses to real budget decisions.

Public Health and Epidemiology

Line charts tracking case counts, hospitalizations, and vaccination rates became a defining visual of the COVID-19 pandemic, and the smoothing debates discussed earlier in this guide directly shaped how public health agencies like the CDC and the UK’s NHS communicated outbreak trajectories to the public. Histograms of age distribution among cases and scatter plots relating population density to transmission rate were both standard tools in epidemiological reporting throughout that period.

Machine Learning and Data Science

Before training any predictive model, data scientists routinely build a full scatterplot matrix and correlation heatmap to understand feature relationships, directly informing decisions in classification analysis and broader machine learning workflows. Model performance is itself visualized, through confusion matrix heatmaps, ROC curves, and residual scatter plots, making visualization as central to model evaluation as it is to initial data exploration.

Academic Research and Peer Review

Journals across biology, medicine, and the social sciences increasingly enforce visualization standards during peer review, requiring authors to show raw data points rather than summary bars alone wherever sample sizes are small. The Springer Nature guide to data visualization for people analytics contexts specifically catalogs tables, heatmaps, scatterplots, line graphs, slopegraphs, waterfall charts, and Sankey diagrams as the standard repertoire expected in professional and academic reporting today.

Higher Education and Student Assignments

Data visualization appears across disciplines far beyond statistics departments, from economics coursework analyzing probability distributions to psychology research reporting survey results, to business case studies built around descriptive statistics. Students preparing a research poster or thesis chapter benefit from the same chart-selection discipline used by professional analysts, choosing the chart type that matches the data structure rather than defaulting to whatever their spreadsheet software suggests first.

How to Create an Effective Data Visualization: Step-by-Step

Building a chart that communicates accurately is a repeatable process, not a matter of artistic instinct. The five steps below apply equally to a classroom assignment, a research paper figure, or a business dashboard.

Students building their first professional-quality figures for a lab report, thesis, or class project can find additional formatting guidance in creating professional charts and graphs for assignments, which walks through software-specific steps for Excel, Python, and R.

Frequently Asked Questions About Data Visualization Techniques