Data Viz for Dummies II
The second unit · more shapes, same film room
More questions, more ways to compare.
The first guide introduced bars, lines, scatterplots, histograms, Sankeys, and treemaps. This second guide keeps the fictional sports setting and asks harder questions: what makes up the total, where do participants leave, and who exchanges with whom? Each section introduces a chart, explains its marks, and names the comparison it helps you make.
Roster check · seven specialists
A chart earns the call-up by matching the structure of the question.
These charts emphasize composition, distributions, losses between stages, exchange or network structure. Each is useful for a different reason; the examples show when that difference matters.
Complex-looking data does not automatically require a complex-looking chart. Start with the plainest view that preserves the claim; call a specialist when the simpler chart drops something essential.
Interactive second-unit board
Scouted by the job to be done
Toggle the lens. The chart names stay put; what changes is the reason each one deserves a roster spot.Stacked bar
compare totals and sharescategories + components
Watch: tiny middle segments are hard to compareArea
emphasize volume over timeordered values
Watch: the baseline becomes part of the claimBox plot
compare spread and outliersgroups of observations
Watch: the raw distribution is summarized awayHeatmap
find clusters and hot spotstwo dimensions + magnitude
Watch: color is poor at exact lookupFunnel
locate drop-offordered stages
Watch: stages must describe one cohortChord
inspect reciprocal exchangesmall many-to-many network
Watch: ribbons become spaghetti quicklyForce graph
discover topology and clustersnodes + relationships
Watch: position is emergent, not a metricStacked bar · combo guard
The total has a group chat, and every segment brought receipts.
A stacked bar carries two readings at once: overall magnitude and the parts that compose it. Normalize every bar to 100% when the mix matters more than the crowd size; keep raw totals when both scale and composition deserve the floor.
Only the first segment shares a baseline across bars. For precise comparisons of every component, try grouped bars or small multiples.
Six home games · audience mix
The opponent changes who fills the building
Every game sums to 100%. Pilots night leans on members; Foxes night recruits the student section.Area chart · power forward
A line learned to occupy space, so the baseline joined the argument.
Area gives time-series magnitude visual weight. It is useful when the quantity itself—attendance, inventory, volume, demand—matters more than tiny point-to-point differences. The filled region makes sustained scale feel sustained.
Because area is measured to a baseline, a cropped y-axis overstates the drama. Multiple opaque areas also hide one another; use stacking only when the series form a meaningful total.
Twelve-game season · building attendance
A wobbling climb becomes a genuinely fuller arena
The fill makes the season-long rise legible as volume, while points preserve the fact that these are twelve measured games—not a continuous sensor feed.Box plot · defensive specialist
The median is only one player; the box plot brings the entire rotation.
Box plots compare medians, the middle half of values, whiskers and outliers in a small space. They are useful for comparing several groups on the same scale.
A box summarizes the distribution but hides its shape and sample size. Add the individual points for a small dataset, or use a violin plot to show where values cluster.
Concession waits · twelve checks per stand
East pizza is steady; north grill has one cursed possession
Compare the medians, then the boxes and whiskers. The north grill’s 17-minute wait stands apart from its other observations.Heatmap · zone defense
When two categorical axes collide, color patrols the entire floor.
A heatmap uses color to show values across rows and columns. Look for a busy time across several zones or a single zone that stays busy. Row and column order can make those patterns easier to see.
Color supports pattern detection, not precision. Include a legend, use a perceptually ordered scale, and do not let a rainbow manufacture boundaries the data never had.
Arena traffic · zone × game moment
Halftime is the boss battle, especially in the south
Every zone is busy in the fifth time slot. The south zone reaches 92, making it a useful place to investigate staffing needs.Funnel · transition tracker
Every stage asks the same cohort: who is still with us?
Funnels make sequential loss explicit. They fit conversion paths where each stage is a subset of the previous one: saw, visited, selected, purchased, attended. The steepest narrowing points to the next question, not automatically the next fix.
Never funnel unrelated stage totals into a fake journey. If people can skip, re-enter, branch, or loop, use a Sankey or process view that admits the truth.
Ticket journey · one campaign cohort
The leakiest handoff happens before seat selection
Of 2,400 fans who saw the offer, 1,460 visited and 910 picked seats. The chart locates loss; research must explain it.Chord diagram · passing savant
A circle of categories starts swapping volume across the lane.
Chord diagrams summarize many-to-many exchange among a small cast. Arc size shows each category’s total involvement; ribbons reveal who trades with whom. They favor the overview—dominant participants, reciprocal pairs, surprising connections—over exact route lookup.
Chord diagrams get crowded quickly. If routes are hard to trace, show fewer categories or switch to an adjacency matrix.
Arena movement · observed zone transitions
Seats and concourse run the motion offense
The thickest two-way relationship joins seats and concourse. Food receives a strong concourse stream, then returns much of it toward the seats.Force-directed graph · connective tissue
Relationships pull together; clusters emerge without assigned seats.
A force-directed graph places connected nodes near one another. Dense groups, bridges and isolated nodes can become easier to spot when you do not already have a hierarchy to organize them.
The layout calculates positions from connections. Distance and direction on the page have no separate measured meaning.
Game-night staff · handoff network
Game ops is the hub; broadcast and guest care bridge worlds
Colors identify departments. Connections across colors show the roles that coordinate work between them.Film room · matchup notes
A specialist earns its place by answering a narrow question.
Choose the form that makes the needed comparison clear, and explain any unfamiliar marks.
Compare a changing mix
Stacked barNormalize for share; retain totals when scale matters.
Feel sustained magnitude
AreaUse a meaningful baseline and restrained overlap.
Compare group spread
Box plotReport n; show raw points when the sample is small.
Find a two-way pattern
HeatmapOrder the axes and provide the color scale.
Locate stage loss
FunnelOne cohort, ordered subsets, no secret re-entry.
Survey reciprocal flow
ChordKeep the cast small and aggregate ruthlessly.
Explore relationships
Force graphPosition is simulated; links are the evidence.
Need exact values
Table firstA chart may accompany lookup; it cannot replace it.
A specialized chart helps when it makes an important feature easier to see.