Data Visualization Operations Flashcards

Question 1

What is a visual encoding in data visualization?

Accepted Answer

A visual encoding is a mapping from data values to visual properties such as position, length, color, size, or shape. Each variable you display is assigned one such channel, so viewers can decode the value by perceiving that property. Good operations start by deciding which channel carries which variable.

Question 2

What is the difference between categorical and quantitative data?

Accepted Answer

Categorical data places observations into named groups without a meaningful numeric order, such as region or product type. Quantitative data consists of numeric measurements where arithmetic and distances between values make sense, such as sales totals or temperatures. The data type determines which encodings and chart types are appropriate.

Question 3

Which visual encodings let viewers compare quantities most accurately?

Accepted Answer

Position along a common scale is generally the most accurate encoding for quantitative comparison, followed by length and angle. Area, volume, and color saturation are harder for people to judge precisely. When precision matters, prefer bars or dots on aligned scales over pie slices or bubble sizes.

Question 4

What does the data-ink ratio describe, and why do practitioners use it?

Accepted Answer

The data-ink ratio is the proportion of a graphic's ink that presents actual data compared with the total ink used. Practitioners use it as a guideline to remove nonessential decoration such as heavy gridlines, redundant labels, and ornamental effects. The aim is a chart where most marks carry information.

Question 5

What is the purpose of a visualization's title and axis labels?

Accepted Answer

Titles and axis labels tell viewers what the chart shows, in what units, and over what range or period. A clear title usually states the finding or the measured subject, while axis labels name each variable and its unit. Without them, even accurate encodings can be misread or ignored.

Question 6

What is a dimension, and how does it differ from a measure?

Accepted Answer

A dimension is a descriptive attribute used to group, filter, or label records, such as date, country, or customer name. A measure is a numeric value that is aggregated, such as revenue, count, or average delivery time. Many tools use this distinction to decide default encodings and aggregation behavior.

Question 7

What is the difference between a continuous and a discrete variable in charting?

Accepted Answer

A continuous variable can take any value within a range, such as height or time elapsed, and is typically drawn as a line or area. A discrete variable takes separate, countable values, such as number of orders, and is drawn as distinct bars or points. Matching the chart to this distinction avoids misleading implications.

Question 8

What distinguishes exploratory visualization from explanatory visualization in practice?

Accepted Answer

Exploratory visualization helps the analyst discover patterns, outliers, and questions, so it is often rough, fast, and interactive. Explanatory visualization communicates a specific finding to an audience, so it is polished, labeled, and guided toward one conclusion. The same data may need both, but the design goals differ.

Question 9

How do you choose a chart type when the main goal is comparing categories?

Accepted Answer

For comparing categories, use a bar chart, with bars ordered by value unless the categories have a natural sequence. Horizontal bars suit long category names, and bars should start at zero so length remains an honest comparison. Avoid pie charts when there are many categories or when slices are similar in size.

Question 10

Which chart type best shows a trend of a measure over time?

Accepted Answer

A line chart is the best choice for showing a trend over time, because connected points make direction and rate of change easy to see. Time belongs on the horizontal axis in chronological order. A column chart suits a few independent periods.

Question 11

When should you use a scatter plot instead of a line chart?

Accepted Answer

Use a scatter plot when both variables are numeric and you want to examine the relationship between them, such as whether higher advertising spend goes with higher sales. Each point represents one observation. A line chart is better when the horizontal axis is time and the sequence of values matters.

Question 12

What chart type helps show how a whole divides into parts?

Accepted Answer

A stacked bar or stacked column shows how a whole divides into parts, with each segment sized by its share. Use a part-to-whole chart only when there are few parts and the comparison that matters is between the total and its components. For many parts, a sorted bar chart often reads more clearly.

Question 13

How is a histogram different from a bar chart?

Accepted Answer

A histogram displays the distribution of one numeric variable by grouping values into adjacent bins, with bars touching to show continuous ranges. A bar chart compares separate categories, and its bars usually have gaps between them. Histograms reveal shape, spread, and skew, while bar charts reveal differences between groups.

Question 14

When is a heat map a sensible choice for a dataset?

Accepted Answer

A heat map suits a grid of values with two categorical dimensions, such as hours of the day by weekday, where color shows the magnitude in each cell. It reveals clusters quickly. Because color is a weaker encoding than position, use a heat map to find patterns, not to read exact values.

Question 15

What makes a pie chart a weak choice for precise comparison?

Accepted Answer

Viewers judge angles and areas less accurately than lengths, so similar slices are hard to rank. Pie charts also become cluttered with many slices, and they cannot show change over time well. If the goal is comparison, a bar chart nearly always does the same job more accurately.

Question 16

How do you decide between a bubble chart and a scatter plot?

Accepted Answer

Choose a bubble chart when a third numeric variable matters and can be encoded as circle size, and a plain scatter plot when only two numeric variables need showing. Bubble size is harder to judge than position, so keep the size range modest and label key points. Adding too many bubbles hides overlapping values.

Question 17

Why should you aggregate data before charting very large raw datasets?

Accepted Answer

Aggregation summarizes many records into totals, averages, or counts, so the chart shows a readable pattern instead of an unreadable cloud of marks. It also reduces rendering time and avoids overplotting. Always state the aggregation method in the chart, because a mean and a median can tell different stories.

Question 18

What is the main risk of plotting averages without showing variation?

Accepted Answer

The main risk is hiding the spread of the data, so two groups with identical means can look the same even though one has far more extreme values. Pair averages with ranges, interquartile ranges, or individual points where practical. Otherwise the chart may encourage conclusions the underlying data does not support.

Question 19

How do you handle missing values before building a visualization?

Accepted Answer

First determine why values are missing, since random gaps and systematic gaps call for different treatment. Then either exclude those records, flag them visibly, or impute them with a documented method. Never silently fill gaps with zero, because a zero looks like a real measurement and distorts averages and trends.

Question 20

Why do you need consistent units and time zones across a combined dataset?

Accepted Answer

Mixing units, such as kilometers and miles, or timestamps recorded in different time zones creates false differences that look like real patterns. Standardize units and convert timestamps to one reference zone before joining or aggregating. Record the conversions in your documentation so others can reproduce the chart.

Question 21

How should you treat extreme values or outliers before creating a chart?

Accepted Answer

First verify whether each extreme value is a data error or a genuine observation, since the answer determines what to do. Genuine extremes should usually stay visible, perhaps with a log scale or a separate annotation. Removing or capping values without disclosure can make a chart misrepresent the process being measured.

Question 22

Why is it important to check whether a join duplicated rows before charting?

Accepted Answer

A join that matches one record to several others can silently inflate totals, counts, and averages, so the chart looks larger or more confident than the data justifies. Compare row counts before and after the join and confirm the key is unique where you expect it to be. Small checks prevent large misreadings.

Question 23

What does it mean to pivot or reshape data from wide to long format?

Accepted Answer

Reshaping from wide to long format turns several similar columns, such as one column per year, into a pair of columns holding a label and a value. Many tools need long format so that a single encoding can map the category and the measure. Reshape before building the chart to keep the process repeatable.

Question 24

How can you keep a data preparation pipeline reproducible for repeated charts?

Accepted Answer

Write preparation steps as code or documented queries rather than manual spreadsheet edits, so every run follows the same logic. Keep raw data unchanged, and store the transformed output separately with a timestamp and version note. Test key totals against a trusted source whenever the pipeline runs.

Question 25

What does it mean to sort categories by value rather than alphabetically?

Accepted Answer

Sorting by value places the largest or smallest categories where comparison is easiest, making the ranking and gaps immediately visible. Alphabetical order suits reference lookups, but it hides the story in most comparison charts. Sort by value unless the categories have a natural order, such as months or age bands.

Question 26

How many colors should a categorical chart use for clarity?

Accepted Answer

Use only as many colors as the message needs, and avoid more hues than viewers can reliably tell apart. For many categories, highlight the one or two that matter in a strong color and render the rest in neutral gray. Limiting palette size reduces confusion and makes legends easier to scan.

Question 27

Why should you avoid using red and green as the only way to distinguish categories?

Accepted Answer

Many people with color vision deficiency have difficulty telling red from green, so those encodings can become indistinguishable for them. Add a second channel such as shape, pattern, direct labels, or lightness contrast. This also helps when a chart is printed in grayscale.

Question 28

What is the role of preattentive attributes in chart design?

Accepted Answer

Preattentive attributes, such as hue, size, orientation, and motion, are features people notice within a fraction of a second without deliberate search. Designers use them to direct attention to the most important element, such as one highlighted bar among many gray ones. Use them sparingly, since too many competing highlights cancel each other out.

Question 29

How should axis scales be chosen for bar charts and line charts?

Accepted Answer

Bar charts should begin at zero, because bar length encodes magnitude and a truncated baseline exaggerates differences. Line charts do not need a zero baseline, since position rather than length carries the value, so a tighter range can show variation clearly. In both cases, label the scale and avoid misleading axis ranges.

Question 30

What does high contrast between text and background accomplish in a chart?

Accepted Answer

High contrast makes labels, titles, and axis text readable at a glance and on varied screens, projectors, and printers. Low contrast, such as light gray text on white, is a common reason that charts fail accessibility checks. Check contrast ratios with a tool and keep text clearly distinct from its background in lightness.

Question 31

Why is chart junk considered harmful, and what counts as chart junk?

Accepted Answer

Chart junk is decoration that adds visual noise without adding information, such as 3D effects, background images, or unnecessary gridlines. It is considered harmful because it competes with the data for attention and can distort perception. Removing it usually makes the remaining data clearer and easier to compare.

Question 32

How do you use annotation to guide a viewer toward the key finding?

Accepted Answer

Add a short text callout directly on or near the relevant mark, stating what the viewer should notice and why it matters. Use a single annotation for the main point rather than many competing notes. Good annotations often sit beside the data they describe, which reduces the effort of matching labels to marks.

Question 33

What is a dashboard, and what separates a good one from a cluttered one?

Accepted Answer

A dashboard arranges related views on one screen to monitor a process or answer recurring questions. A good dashboard puts the most important metrics first, uses consistent scales and colors, and limits itself to a small number of views. A cluttered one shows everything and leaves viewers to search for answers.

Question 34

What are filters and parameters, and how do they differ?

Accepted Answer

Filters limit which records are included in the view, such as showing only one region or a date range. Parameters are user-chosen values that change a calculation or threshold, such as a target rate or a forecast horizon. Filters change what data is seen, while parameters change how values are computed or compared.

Question 35

Why should a dashboard display the time period and data refresh date?

Accepted Answer

Viewers need to know what period the numbers cover and how current they are, or they may act on stale or mismatched figures. A visible refresh timestamp builds trust and helps people notice when a data feed has stopped updating. Place this information near the title where it is hard to miss.

Question 36

What is drill-down interaction, and when is it useful?

Accepted Answer

Drill-down lets a viewer click a summary mark, such as a region total, to reveal the underlying detail, such as stores or transactions within that region. It is useful when a summary raises a question that a single static view cannot answer. Keep the starting view understandable on its own.

Question 37

How do you make a dashboard accessible to people using screen readers or keyboards?

Accepted Answer

Provide text alternatives or descriptive summaries for charts, so screen reader users receive the key finding. Make sure every interactive control can be reached and operated with a keyboard, and label each control clearly. Also avoid relying on color alone, and check that focus indicators remain visible.

Question 38

What should you check when a dashboard's underlying data source changes?

Accepted Answer

Confirm that field names, data types, and key totals still match what the dashboard expects, because a renamed column can silently break calculations. Compare recent figures with the source system to catch changed definitions or duplicated loads. Then update documentation so that calculated fields and filters reflect the new structure.

Question 39

How do you decide which metrics belong on a dashboard's main view?

Accepted Answer

Start with the decisions or questions the audience must answer, then select the few metrics that directly inform those decisions. Each metric should have a clear definition, an owner, and a target or comparison point where possible. Move supporting detail to secondary views so the main view stays focused.

Question 40

What is the risk of letting many users edit a shared dashboard without governance?

Accepted Answer

Without governance, metrics get redefined in slightly different ways, so two dashboards may report different numbers for the same concept. Shared dashboards also accumulate abandoned views and duplicate filters that confuse new viewers. Establish naming conventions, a review step for changes, and a designated owner for each dashboard.

Data Visualization Operations Flashcards

1Core visualization concepts

2Selecting appropriate chart types

3Preparing data for visualization

4Design, color, and perception

5Dashboards, interaction, and upkeep

Learning Resources