Data Visualization: Taxonomy and Tools

Data Visualization: Taxonomy and Tools

4.1 Where We Are and Where We Are Going

4.1.1 The Plan for Today

This session closes out the taxonomy topic and opens the door to the tools of the trade. The agenda runs like this: a quick recap of the last session (storytelling and the taxonomy of charts), a refresh of the fundamental building blocks — data types, relationships, and complexity — because they are not part of the course handout but they decide which graph you reach for, then a few short live demos comparing graphs on real data, a small quiz of four or five questions to lock it in, and finally, in the second half, a survey of what data visualization tools exist in the market today, including a look at the Gartner quadrant, which classifies tools as emerging, gaining popularity, or leaders.

The tools topic was pulled forward from the next session on purpose. The taxonomy material is light and nobody benefits from staring at slides for two hours, so the plan is: see the whole tool landscape today, go deeper into individual tools in the next session (desktop-based tools, online visualization tools, their libraries, and open source versus proprietary options), and from the sixth session onward, zoom into Tableau desktop with live demos and hands-on work.

Think of the arc this way: the taxonomy is the theory of choosing (which chart for which job), today is the market tour (what tools exist), and the coming sessions are the hands-on practice (building charts yourself). Each stage builds on the previous one, so the recap matters — the building blocks we refresh now are the ones every chart decision in the demos will lean on.

4.1.2 Recap: Storytelling and Chart Taxonomy

Last session we covered storytelling and its three sections. First comes the setup, where we start and give the context. Then the conflict, where the challenges are discussed. Then the agreement, where the story ends with actions and what we will do next. We also went through the do's and don'ts of storytelling.

From there we started the taxonomy of data visualization — the approach of classifying and organizing different types of charts: in what scenario, which chart usually does better. There were chart suggestions for each scenario. We looked at a few charts in detail — the dot plot, the bar chart, and the floating chart (a floating chart is one not tied to an axis baseline; it floats). For each chart we noted the pros, the cons, and the scenarios where it fits. The course material for this section is available in the file section if you want to revisit it.

Two memory hooks for the recap:

• Bar chart — a bar starts at zero on the axis, so its length encodes the value directly. If you want exact magnitudes, this is the workhorse.
• Floating chart — a bar (or line segment) that does not start at zero; it stretches from a low value to a high value. Because it floats, it shows a range (like a project span from start date to end date) rather than a single total. The trade-off: you lose the zero baseline, so you lose quick absolute comparisons.

That contrast — length from a common baseline versus a floating range — is a preview of a theme running through the whole taxonomy: every chart choice trades some kind of precision for some kind of clarity, and the right pick depends on the message.

4.1.3 Getting Tableau Desktop

Before the recap began, there was a practical check on tool access. The institute has an analytics lab, and students can register there for a Tableau Desktop license. One student had already registered and received an email with a product key, though installation was still pending. The key opens the desktop version free for one year. When you launch the product (it arrives as a typical executable), it asks for the key, and for a year you can use all of the features on your own. The key expires after a year and can be renewed. If you are stuck, reach out to the batchmates who already have it.

Practical notes worth keeping:

• The license is free for one year from the moment you activate it, covering every feature of the desktop product.
• When the year ends, the key expires and can be renewed — renewal was confirmed to work, so plan to renew rather than pay.
• The class is a support network: if you cannot install or activate, ask batchmates who already did it. Do not sit stuck on a tool-access problem; it is a solved problem for half the batch.

This session's live demos all run inside Tableau Desktop, so having it installed before the next session means you can follow along hands-on instead of just watching.

4.1.4 What This Session Builds Toward

Keep the session map in mind as you read on: the first half (sections 4.2–4.12) rebuilds the decision framework — data types, relationships, complexity — and tests it with demos and a quiz. The second half (sections 4.13–4.14) tours the tool market and gives you criteria for choosing a tool. The quiz in section 4.12 is not a trap; it is the study material — each question's reasoning shows you exactly how the framework turns into a chart choice.

4.2 Data Types: The Four Building Blocks

4.2.1 Categorical Data

Categorical data is data made of distinct categories — labels rather than numbers. A chess game, for example, has moves with statuses; the statuses are labels, and then there are counts around those labels. "Small, medium, large" are labels. Movie genres are labels. Anything category-based with labels counts as categorical data.

In the reference literature this kind of data is often called categorical-nominal (nominal means "named"): the values name a group — country, gender, movie genre — and the names carry no number at all. Think of a dataset of movies: genre is categorical (Drama, Comedy, Sci-Fi), studio is categorical, but box office revenue is not.

Two properties matter. First, there is no required order — the sequence of the categories does not matter, so the data need not be sorted. A bar chart with "Comedy" on the left and "Drama" on the right tells the same story either way; sorting is a choice you make to help the eye, not a requirement of the data. Second, because the number of labels is usually small, categorical data is typically visualized with bar charts, pie charts, or heat maps. A pie chart can handle a small number of labels well.

There is a catch worth repeating: pie charts should be used sparingly when there are a lot of data points. Our brain struggles with angular formatting — it cannot easily compare magnitudes by looking at slice angles. A pie chart works when you are visualizing a whole: out of a hundred percent, how much of it belongs to each value. For anything with too much context, a bar chart or another graph serves better.

4.2.2 Numerical Data: Discrete and Continuous

Numerical data is data that can be measured — straightforward. The one level of detail worth keeping is that numerical data splits into two kinds. Discrete data is whole numbers — counts like the number of people or the number of cars. Continuous data can take any value within a range, including decimals — a share price or a temperature, for example.

A clean way to separate them: counts are discrete, measurements are continuous. You can have 3 people in a room or 4, but never 3.5 people; you can have a temperature of 21.7 °C or 21.73 °C. The reference material draws the same line with the terms interval-scale (dates, temperature — equal gaps, no true zero) and ratio-scale (prices, age, distance — has a meaningful zero, so ratios like "twice as old" make sense). The professor's discrete/continuous split is the everyday version of this distinction, and it is the one you will actually click in a tool.

This distinction is not just theory. In Tableau — and in any data visualization tool worth its name — when you select a numerical field you get the discrete and continuous options, and you decide in what form to show the value. You are telling the tool how to treat the field.

4.2.3 Ordinal Data

Ordinal data looks like categorical data, but with a key difference: it has a sequence, an order — that is why it is called "ordinal." The order only makes sense when you sort the values. Surveys are the classic example: "strongly agree, agree, neutral, disagree, strongly disagree" sits on an underlying scale, and the sequence matters. Because the ordering matters, ordinal data is usually visualized with a bar chart or a box plot.

The reference taxonomy calls this categorical-ordinal, and the classic example is Olympic medals: gold, silver, bronze. They are labels (you cannot compute "gold + silver"), but their order is fixed — bronze is below silver, silver below gold. Likert-scale survey responses ("strongly agree" through "strongly disagree") work the same way.

The one practical trap: ordinal data looks categorical, so it is easy to treat it as nominal — and then the tool sorts the labels alphabetically, and your carefully ordered scale ("strongly agree" on the left, "strongly disagree" on the right) gets scrambled into "agree, disagree, neutral, strongly agree, strongly disagree." Keep the field's order explicit, and sort the axis in the intended sequence before presenting.

4.2.4 Time Series Data

Time series data is anything where each data point is associated with a specific time. Stock price data is the classical example. Tracking your website traffic by hour is another. Temperature recorded hour by hour or minute by minute — the way weather apps show a day's temperature — is a third.

For time series, line graphs usually work best, although heat maps are also used. Typically you will see a date-time value on the X axis to make the chart meaningful. Tableau automatically categorizes a date field into years and quarters, and you can opt for the exact date instead. The same data can be drawn as a bar chart, a line chart, or a year-wise heat map — the tool leaves the choice to you, and you choose based on what you want to show. When the data is too much, the line graph is the safer default, and you can always undo and try another form.

4.2.5 Why Data Types Matter

The whole idea of reviewing data types is that the data type, seen through the X axis and the Y axis, tells you which graph to choose. With time and experience this becomes second nature, but at the very least you should know fundamentally which type of graph goes with which data type. That is the entire point: categorical needs no sequence; numerical is countable and can be discrete or continuous; ordinal is categorical with sequenced labels; time series has time attached to every point.

4.3 Relationships in Data Visualization

Once you know the data types, the next question is: what relationship are you trying to show? Four relationship families cover most chart decisions.

The four families — comparing magnitudes, trends over time, distributions, and correlations with networks — are deliberately the same shape as the data-type building blocks from section 4.2: the data type tells you what you can plot, the relationship tells you what you should plot.

4.3.1 Comparing Magnitudes

The first relationship is comparing the magnitude — a quantitative comparison between data points. You compare sales figures between one region and another, or sales of one product against another. The focus is on the numbers themselves.

Bar charts work well here, and pie charts work too, but with a specific job: a pie chart is a whole-to-part view. If a budget is 1 million, a pie chart shows how that 1 million is broken up across the different items — proportional contribution. A stacked bar can do the same job. But when you need the exact magnitude, the bar chart wins, because it tells you the number directly while a pie chart does not give you the exact magnitude. Use pie charts for proportional contributions, not for exact magnitudes.

4.3.2 Trends over Time

The second relationship is trends — changes or patterns over time, exactly the time series case from the data types. Line graphs are the standard choice. An area chart is a line chart with the space under the line shaded — useful for the same trend story. A scatter plot also looks at a relationship between two things, but it need not be date-wise; it compares two variables directly. Whenever you are showing trends, these are the visualizations to consider.

The scatter plot belongs to the same neighborhood but does a different job: it compares two variables against each other (say, price against volume) with no time requirement, so each point is a pair of values. You will meet it again in the correlation family (section 4.3.4) — a reminder that chart families overlap, and the primary message decides the home.

4.3.3 Distributions

The third relationship is distribution — the spread of values. A typical example: marks in a class. There is a spread showing how many students scored at each level. You have seen this in every organization's year-end appraisal, where management builds a bell curve and places most people in the good or strong categories, with outliers and exceptional performances at the ends. That is a distribution.

For distributions you use histograms, box plots, or density plots. This relationship family is extremely common in data visualization.

4.3.4 Correlations and Networks

The fourth relationship is correlation, and with it networks. The typical graphs are heat maps, correlation charts, and network diagrams, which show the connection between one point and another. These are not the most common charts, but they are not rare either — heat maps and correlation charts are quite common. From there, the choice depends on the data, the use case, and the business case.

A Sankey diagram belongs to this family. Sankey diagrams show flow: how things flow from one stage to another and what they ultimately lead to. A classic example is immigrant flow — how many people moved where over a period. Another is a soccer world cup: all the countries, then how the teams reached the last eleven, then the semi-finals, then the final — the whole tournament as a flow. A third example breaks a country's GDP down into how much went to healthcare and the other categories. Sankey diagrams are heavily used in supply chain contexts. They are not everyday charts, and your business case may never require one, but keep them in mind — whenever you are showing any relationship, these are the possible visualizations, and the correct one depends on the data and the use case.

4.4 Complexity: Simple, Multi-variable, Interactive

The third dimension of the recap is the complexity of the visualization — three levels running from simple through multi-variable to interactive.

4.4.1 Simple Visualizations

Simple visualizations are quick and clear — they communicate a single message, sometimes just one number or one graph. Bar charts and line graphs are the workhorses here. They are great for showing trends and comparisons, and they are easy to understand because the audience has been seeing them since school; people do not have to think too much. From the stakeholder's point of view, these are the simple class.

4.4.2 Multi-variable Visualizations

Multi-variable visualizations come in whenever you compare two or more variables. This is a slightly deeper analysis than the simple class. These charts can reveal complex relationships — how different values change with the variables — and they require a little more interpretation from the audience. It is not like looking at a bar chart where you immediately see that sales were high this month and low that month. But these are still very commonly used.

The rule of thumb: if the message is one comparison → simple. If the message is how two things interact → multi-variable. The moment the audience has to work to find the message, you have probably crossed a complexity level.

4.4.3 Interactive Visualizations

Interactive visualizations are what you have seen in dashboards: filters and criteria that let users pick and choose what they look at. Interactivity helps users explore hidden insights, and it helps stakeholders get familiar with the data faster.

4.4.4 The "Start Simple" Principle

A guiding rule applies to any kind of data visualization: start simple. Do not load the chart with everything at once — start simple, keep adding the other stuff only if needed. Keep the communication clear. Support the data and the numbers with commentary — a chart of numbers without explanation does half the job. And give the interactivity, so users can pick and choose, drag and drop, and filter on their own.

The complexity dial connects forward to section 4.5: the bar-versus-pie live comparison is deliberately a simple chart demo — the professor's point is that even at the simplest level, chart choice changes what the audience can extract.

4.5 Bar Chart versus Pie Chart: A Live Comparison

4.5.1 The Setup

To show why the bar chart beats the pie chart for magnitudes, we looked at a very basic dataset — just two columns — and rendered it both ways. No rocket science, but the contrast is instructive.

4.5.2 What the Bar Chart Gives You

With the bar chart, you can read off the exact magnitude. You can even switch the labels on, and the chart tells you directly: the value here is five, the value there is seven. The magnitude is visible in the bar length, and the number is on the chart — no guessing.

4.5.3 What the Pie Chart Hides

With the pie chart, you struggle. Is that slice really 12? At first glance you cannot make out the exact count at all. The pie tells you that this size is bigger than that one — relational size — but beyond that, exact magnitude mapping is exactly what the pie chart cannot give you.

4.5.4 The Takeaway

So the rule holds: pie charts help with proportional contribution, while bar charts are better for precise comparison. When it comes to comparing magnitude, bar charts are preferred because the comparison is easy and exact. This is the same point that came up in the data-types recap — the brain simply cannot extract exact numbers from angles, and even a quick glance at a bar chart gives you the magnitude, with or without labels.

This is one of the rare data visualization rules that holds across every domain — it is the same reasoning financial analysts apply when comparing quarterly revenue and the same reasoning supply-chain teams apply when comparing warehouse volumes. The chart does not change the number; it changes how accurately the audience can read the number.

4.6 Line Charts, Area Charts, and Heat Maps for Trends

4.6.1 The Line Chart Walkthrough

The trend demo used stock price data. Put trade date in the columns shelf and close price as the value, and Tableau drew a line graph automatically — the data had few rows, so a line came up immediately. Zooming in shows the values for different days. The default view aggregated to year or quarter, but the tool offers an exact date option, and you toggle it to see the real granularity. With a longer dataset, the overall trend becomes visible at a glance: prices going up, prices going down, the whole shape of the market.

The same data can render as an area chart to show the trend — the shaded line — but the plain line graph is the easier default. You will learn this in the hands-on sessions: you can add two parameters at once. For instance, show the close price as a line and the volume as bars in the same view, and select the properties of each chart individually.

4.6.2 Heat Maps for a Yearly Comparison

The heat map is not ruled out for time series. With the same stock data, Tableau produced a year-wise heat map comparing 2021 and 2022 in one glance — this is the 2021 sum versus the 2022 sum, per period. If you are showing something at a yearly level where the data is small, a heat map works fine. If the data is dense, the line graph stays the better choice — and you can always undo.

Real-world connection: this exact pattern — line for the trend, volume bars underneath, yearly heat map for comparisons — is the standard toolkit in stock-market dashboards and portfolio reports, where analysts must distinguish a genuine market move from a thin one in seconds.

4.7 Histograms for Distributions

4.7.1 The Histogram Walkthrough

The distribution demo used a sample of grades. Select the test score field and choose the histogram option, and Tableau shows the distribution automatically — how many students scored in each category. The tool creates the bins on its own; the course material covers in detail how bins are created and what you can tune. A histogram is a fast, easily understandable way to showcase bins and the values that fall into them.

4.7.2 Histogram versus Box Plot

Both histogram and box plot reveal distribution. The box plot shows quartiles — the top quartile and the rest — but for keeping it simple, the histogram is the better choice for the distribution of any parameter. That distinction came up again in the quiz below.

4.8 There Is No Single Correct Taxonomy

4.8.1 The Chart Choice Belongs to the Audience

A summary point, repeated because it matters: there is no single correct taxonomy for data visualization. We choose what is better for the situation — specifically, what the audience wants to see. If the audience wants detail, we choose one graph; if the audience is more high-level, we choose a very different one. There is no fundamental formula.

The practical consequence: when you select a chart for a stakeholder, ask what they will do with it. An executive scanning for direction needs a simple trend line; an analyst hunting for outliers needs a distribution; a regulator checking compliance needs exact numbers. Same data, different audiences, different charts — all correct.

4.8.2 The Fundamentals That Actually Move the Needle

What carries you is clarity on the fundamentals. That means controlling cluttering — not overloading the chart — understanding pre-attentive attributes (the visual properties the eye processes before conscious attention), knowing how to use color, and knowing how to use fonts. If these fundamentals are in place and you support them with the right graph selection, your data visualization and presentation skills will genuinely help your audience understand.

These small things are usually neglected. People do not give importance to them, but that is exactly the difference between a visualization that communicates and one that confuses — and learning to apply the right concepts, the right approaches, and the right chart selections is the whole purpose of the topic.

Real-world connection: this is the difference between dashboard designers whose work ships to executives and those whose work gets ignored — analytics teams that master these fundamentals produce dashboards that change decisions, because the decision-maker sees the message before they even think about the chart.

4.9 Geospatial Data Visualization

4.9.1 How Automatic Mapping Works

Geospatial data — data tied to places — is another case where the tool does the heavy lifting. Tableau is intelligent enough to map geospatial data on its own. Some datasets ship with Tableau; the demo used the world indicators dataset. When a field is geospatial, it carries a globe icon — that icon is the signal that the tool recognizes the field as location data.

4.9.2 The Walkthrough

Place the country field on the view and the tool draws the map automatically. Then you can ask for any parameter: show GDP and the map fills in the GDP numbers for each country; color by birth rate and the map immediately shows which countries have higher birth rates — Africa stands out at a glance. The tool has its own built-in database: the moment you enter a standard place name, it matches the name, looks up the coordinates, and places the point on the map. We do not have to do any of that complexity ourselves.

4.9.3 Data Preparation: Use Standard Names

The one thing we must take care of is the data itself: the country, city, or state names must be the standard names — no abbreviations, no full stops. As long as the naming is standard, the tool's intelligence handles the rest. You can also switch on the labels option and the map shows the name of every location plotted.

4.9.4 What to Expect with Country-Level and State-Level Data

Because the demo dataset was global, the map covered the whole globe. Had the data been India-only with different states, the map would have shown only India and created all the states within it, letting you compare state-wise data. We will see those demos in the subsequent sessions.

Real-world connection: this is how logistics dashboards plot delivery hubs, how healthcare dashboards map outbreak spread by region, and how retail chains review state-wise sales — in every case the analyst's job is preparing standard location names, and the tool's job is drawing the map.

4.10 Radial (Polar) Charts

4.10.1 What They Are

A radial chart — also known as a polar chart — uses a circular axis instead of the typical X-Y axes. You see these more and more in reports and presentations these days, so it is worth knowing them. In one glance you can read the value and the comparison across all series.

4.10.2 The Strengths

First, space: the radial chart takes less room. In one space, the example chart showed five values — a bar chart would have needed far more room for the same information, even though the values are just percentages. Second, appearance: these charts are visually appealing, more attractive and engaging than a plain bar chart. Third, you can show the relationship between the series with it. For these reasons the radial chart is getting very popular, and you will see it in a lot of presentations.

4.10.3 The Limitations

The challenges are real. You cannot compare precise values — a bar graph lets you see exactly how the bars move against the axis, but with a radial chart it is difficult to pin down exact numbers. You cannot show big datasets, or the chart becomes too hard to read. And it is not yet familiar to the average audience — the chart is still gaining popularity, unlike the bar graph people have seen since school. If you use one, expect to explain it with your commentary.

Real-world connection: radial charts are the look of modern analytics dashboards and annual reports — performance scorecards (a finance team scoring each department on the same wheel) and fitness apps (activity by hour of the day, riding the 24-hour circle) both rely on this format. A note from the reference material: radial charts work well for time when the sequence is continuous, like a 24-hour clock.

4.11 Glyph Charts

4.11.1 What They Are

A glyph chart represents data points with symbolic shapes — these symbolic markers are the glyphs. Glyph charts are useful for multivariate data and usually show the relationships between the variables. There are hundreds of chart types out there; you do not need to pressure yourself to know all of them, but this one is worth recognizing.

There are hundreds of chart types out there; you do not need to pressure yourself to know all of them, but this one is worth recognizing.

4.11.2 Customizing Symbols in Tableau

In Tableau you can choose the shape for each series: a square, a diamond, a cross, a star — whatever communicates best. You select the shape per series, so one series is drawn with crosses and another with stars. All of this is available inside the tool. Personally, I have not used glyph charts much, but we can look at them in subsequent sessions. Do not get biased toward one particular chart; even theoretical knowledge of the alternatives is good enough.

4.11.3 When They Help

Glyph charts fit when you have a lot of information to present in a compact space. Looking at too much data at once overwhelms the audience; glyph charts help you engage the end user better. They highlight relationships, they are flexible, and you can show different kinds of pictures and icons.

4.11.4 The Limitations

The disadvantages are significant. There is a learning curve — if the radial chart is not popular, the glyph chart is the next level of unfamiliar. These graphs are not used very commonly, so they require more effort to interpret. You cannot pinpoint the accuracy or the position of the data, and with too much data there will be potential overlaps — the glyphs have a very limited set of positions, so there are scalability issues: as the data grows, interpretation gets harder. They are also not easy to design — the other graphs took seconds to build, but glyph charts take more effort because symbols, color, and size all interact. If you are not careful, you will make interpretation harder for the end user. And there is an accessibility problem: people with color blindness, or people not well versed in reading such charts, may find them a real challenge.

Real-world connection: glyph-like designs appear in the OECD Better Life Index (the flower glyph in the reference material) and in modern sports analytics, where player 'radar' and glyph profiles summarize many performance variables in one compact shape — always for exploration, never for exact reporting.

4.12 Chart Selection Quiz

Four or five questions closed out the taxonomy topic, done together in class — no right or wrong pressure. Each question and its reasoning follows, because the reasoning is the study material.

4.12.1 Best Chart for Comparing Sales across Regions

The reasoning in three steps: the question is a magnitude comparison across categories (regions), so the bar chart family applies. The line graph is rejected because it is mostly for time series — no time axis exists here. The scatter plot is rejected because it is for correlation — no two-variable relationship is being examined. One relationship family, one correct family member.

4.12.2 Best Visualization for the Distribution of Exam Scores

The reasoning in three steps: the word "distribution" names the relationship family directly, and the histogram is that family's straightforward chart. The box plot is the legitimate alternative — it shows quartiles — but it compresses the shape, so for a single parameter the histogram stays the better choice (the same point from section 4.7.2).

4.12.3 Time Series Plot versus Line Graph

The reasoning in three steps: the question is a terminology contrast — "time series plot" versus "line graph" sound alike, so the exam needs the precise distinction: a time series plot is for a single variable over frequent intervals (share price, market value). Multiple variables over time are challenging. Correlation and discrete data points are both outside this question. The precise condition — single variable, frequent intervals — is the deciding detail.

4.12.4 Heat Maps Are Only for Categorical Data

This question carries a misconception correction: the claim "heat map only for categorical data" sounds plausible because heat maps often color categories, but the year-wise stock price demo (section 4.6.2) proved the claim false — numerical time series data renders just as well as a heat map. The trigger is the phrase "only for categorical data"; the resolution is that heat maps work for numerical and time series data too. Remember it as: the heat map's job is value-by-color; the data type is up to you.

4.12.5 Scatter Plot versus Line Graph for Trends

This question carries a second misconception correction: "scatter always beats line for trends" sounds reasonable because scatter plots are sophisticated, but the correction is the reverse — a line graph shows the overall trend better, while a scatter plot shows individual data points. The word "always" is the giveaway: the scatter plot is the right tool for point-level relationships, the line graph for trends. If your goal is the trend, the line graph is the tool.

4.13 Data Visualization Tools in the Market

4.13.1 The Gartner Quadrant

Ask ten people for the top tools and you will get different lists — there is no one definition of the top five or top ten. What gives a more structured view is the Gartner quadrant: Gartner evaluates the data visualization tools every year and classifies them — which tool is emerging, which is gaining popularity, which is the leader — and publishes the classification as a quadrant.

4.13.2 Tableau

Tableau is a very powerful and very versatile data visualization tool — you have already seen a glimpse of it in the demos in this session. It is drag-and-drop driven: simple sheets, dashboards, and storylines all become easy to build. Beginners can use it, and with experience you build even the more complex reports.

A key feature beyond drag and drop is calculated fields: you write your own formulas inside the tool — "if this is greater than that, then…" — so the analysis is not limited to the fields that come with the data. On pricing, Tableau offers a free personal edition — Tableau Public, the web version — and paid plans for enterprise needs, plus student plans and corporate plans, so different users get different options.

4.13.3 Microsoft Power BI

Power BI is very popular and integrates seamlessly with Microsoft products and services. For organizations already on the Microsoft ecosystem — Excel and the rest — Power BI slots right in, which is a clear plus for those companies.

The pricing model was discussed in class because it surprises people. The desktop version — Power BI Desktop — is free for everyone. It includes Power Query, the data preparation tool, just like the one inside Excel. But when you want to collaborate with the team, share with leadership, or present dashboards to others, you need Power BI Service, and that is paid — that is where the cost comes in. Individual usage is allowed for free; sharing and circulation is what triggers the paid plans.

Power BI is very powerful, with a lot of features and drag-and-drop options, and organizations already in the Microsoft ecosystem gain from that alignment. Corporate-wise, Power BI probably has a slight edge; but both tools are equally popular and very widely used.

4.13.4 Looker Studio (formerly Google Data Studio)

Looker Studio — formerly Google Data Studio — is a free and easy-to-use tool from Google. You can create basic reports and dashboards with it, and it is usually used for quick data exploration and visualization. It may not be as popular as the more mature tools, but it is widely used — you see posts and news about it in the media constantly.

When exploring it, you find you can connect Google Sheets and do almost everything the bigger tools do — but it is not as flexible or as powerful as Power BI or Tableau. For early-stage visualization and data exploration it is a good fit. And with the brand backing of Google, more features will keep coming as the competition drives it forward.

4.13.5 Grafana

Grafana is an open source platform, mostly preferred for time series data. It supports DevOps and infrastructure monitoring, and it comes with plenty of charts and plugins. It is quite powerful for its niche, though when it comes to the vast feature options and the range of data sources, it is not as versatile or as widely supported as the bigger commercial tools. It is one of the emerging open source tools — good to know about, and worth keeping an eye on. The links to these tools' websites are provided in the course material so you can explore them yourself.

4.13.6 Qlik (Qlik Sense)

While the tools so far are purely visualization tools, some tools combine a BI platform and a data analytics platform — they give much more than simple drag and drop. Qlik is one of those. Qlik Sense is a self-service BI platform: you explore and analyze data without needing technical expertise, and it offers drag and drop — but it is more than drag and drop.

Qlik is a mature product with many offerings: Qlik Cloud for cloud data integration, and Qlik Compose for data lakes and data warehouses. Qlik Sense itself also comes as an on-premises solution for highly regulated industries. That breadth is exactly why Qlik sits in the Gartner leader category.

4.13.7 Sisense

Sisense is another BI platform, quite common. It does in-memory analytics and handles very large datasets, with a very user-friendly interface and a lot of data connectors. It provides APIs and SDKs, so you can integrate its dashboards with your own data sources and write your own functions. It supports many industries and covers cloud, data connectors, and data visualization. It is a mature product and it appears in the Gartner quadrant.

4.13.8 Domo

Domo is a cloud-based BI platform with a lot of visualization options — BI, analytics, and a data foundation among its features. It is fast gaining popularity: cloud-based tools are getting easier to use, and with Domo you simply drag and drop. It is an emerging tool that can do almost everything — dashboards, reports, and very polished visualizations. There are a few versions (the details and pricing were not covered in depth), so it likely offers a basic version with enterprise options on top.

4.13.9 Infogram

Some tools are specialized, and Infogram is one of them: it is mainly used for creating infographics — where you combine graphics and information. It is very easy to use, with ready-made templates. In the live look at the tool, you pick a template, drag and drop your content, and place a line graph from the left-hand panel — you can create reports in minutes.

It has a collection of designer templates, supports animation and interactive charts with very little effort, and can produce infographics, reports, slides, dashboards, and even maps. It can integrate with social media. A lot of players are emerging in this space, and Infogram is a good one to familiarize yourself with.

4.13.10 ChartBlocks

ChartBlocks lets you create interactive charts with a lot of customization, and you can embed the charts in websites and other applications. It is gaining popularity, with templates and drag-and-drop that make it easier to use than the big platforms. It is not as powerful or mature as Power BI or Tableau, but for small-scale needs — when your requirement does not involve very complex financial models and heavy graphs — these tools offer good, easy options. It will have paid options of its own, but the pricing plan was not covered in depth.

4.13.11 Flourish

Flourish is built for data storytelling: you create interactive data stories with a lot of templates. It is mainly for storytelling — it may not have all the analytical machinery of the big BI tools — but the story-building is excellent. You drag and drop to make bar charts, magic charts, animated 3D graphs, survey data, interactive timelines that change with time automatically, and story maps.

The demo walked through a real story: population by country, then zooming into urban population, then into Asia and Europe, then ordering highest to lowest. The story navigates, the graphs animate with real data — for example, how different countries' GDP moved across the years, with the year changing on screen and the countries moving in the rankings. The example ran until 2016; had the data run until 2014, India would have been on top. Color palettes and visualizations are all handled in an easy, drag-and-drop manner. This is the style of real-time data infographic you see all over social media.

4.14 Choosing a Visualization Tool

4.14.1 The Selection Criteria

Every tool in the market has something going for it — that is why it is in the market. The choice depends on three things: your needs, your budget, and your expertise. One will naturally choose the tool they already have expertise in. Combining all three, you pick the tool that suits you — and that is exactly how a professional selects a tool for a business case.

The tour makes one point: our view is often limited to one or two tools, but look at what exists in industry. Marketing companies and many others need jazzy, powerful visuals, and there are many players emerging in that space.

4.14.2 What Comes Next

The theory phase is done — you have learned the fundamentals. The next session goes deeper into the tools: desktop-based tools, online visualization tools, their libraries, and open source versus proprietary options, plus the architectures behind them. From the sixth session onward, the focus is Tableau desktop with live demos and hands-on work. Whatever form you need, the instruction is to take what you have learned here and use it — you have the fundamentals now, and the tools are the vehicle.

4.15 End-of-Session Discussion

4.15.1 Plans for Quizzes

The answer sets a clear contract for the rest of the course: quizzes are coming, they will be based on what has been learned in class, and the plan will be published in advance. Nothing will arrive unannounced.

4.15.2 Feedback on Ambiguous Quiz Questions

A student raised a fair point about multiple-choice quizzes in general: some questions are ambiguous — both options can be correct. When both options are correct but the answer key marks only one, students lose marks even though their answer was right.

The exchange closed with the acknowledgement that ambiguity in chart-selection questions is a real problem — the same reasoning that runs through the whole taxonomy topic: with charts, several answers can legitimately be right depending on the scenario, so questions and answers must be phrased to respect that.

Exam Guidance Summary

• Quiz plan: Quizzes will be announced and the plan shared in advance — nothing will come as a surprise. The quizzes will be based on what has been learned in class. (Exam note: expect quiz questions modeled on the chart-selection reasoning from section 4.12.)
• Exam question quality: The exam will avoid ambiguous multiple-choice questions. Answers will be clear and unambiguous — especially important in data visualization, which does not have a single right answer like a theorem. If a question genuinely has two correct options, "both are correct" should not cost marks.
• Chart-selection reasoning to study: the quiz logic is the model. Bar chart for comparing magnitudes across categories (line is for time series, scatter for correlation). Histogram for distributions (box plot as the alternative that shows quartiles). Time series for a single variable over frequent intervals. Heat maps work for numerical and time series data, not only categorical. Line graphs beat scatter plots for overall trends.
• Conceptual foundations: know the four data types (categorical, numerical with discrete/continuous, ordinal, time series), the four relationship families (magnitude, trends, distribution, correlation/network), and the three complexity levels (simple, multi-variable, interactive) — these drive every chart choice.
• Chart knowledge: for each chart in the taxonomy — bar, pie, line, area, heat map, histogram, box plot, scatter, Sankey, radial, glyph — know what it is for, its strengths, its limits, and when to pick it (the bar-versus-pie rule, the histogram-versus-box-plot rule, and the audience rule from section 4.8 are the most likely to appear).
• Tool landscape: be able to place the main tools — the Gartner quadrant leaders are Microsoft (Excel and Power BI together), Salesforce, and Tableau–Qlik; know what the emerging tools (Looker Studio, Grafana, Sisense, Domo, Infogram, ChartBlocks, Flourish) are mainly used for.
• Data types → charts mapping (study aid):

Key Industry Applications

• Gartner quadrant: used across the industry to evaluate BI and visualization platforms; current leaders are Microsoft, Salesforce, and Tableau–Qlik. Real-world: companies use this quadrant to shortlist tools for their BI stack.
• Microsoft ecosystem: Power BI Desktop (free, with Power Query for data preparation) plus Power BI Service (paid, for sharing and collaboration) — the standard stack for organizations on Microsoft. Excel remains a widely used data visualization tool, which is why Microsoft's leader position in the quadrant covers both.
• Tableau: dashboards, storylines, and calculated fields in analytics teams; Tableau Public serves as the free personal edition; one-year free licenses available through academic analytics labs.
• Looker Studio (Google Data Studio): free quick exploration and reporting on Google Sheets data, popular in media and small teams.
• Grafana: open source monitoring and dashboards for DevOps and infrastructure teams, optimized for time series data.
• Qlik: Qlik Sense self-service BI for non-technical analysts; Qlik Cloud and Qlik Compose for data integration and data warehouses; on-premises Qlik Sense for highly regulated industries.
• Sisense: in-memory analytics on very large datasets, with APIs and SDKs for embedding dashboards into products.
• Domo: cloud BI for dashboards and reports, gaining ground in enterprises.
• Infogram: infographics, reports, and slides with designer templates, integrated with social media — used by marketing teams.
• ChartBlocks: embeddable interactive charts for websites and applications.
• Flourish: animated, data-driven stories (population and GDP rankings over the years) — the format you see in social media and news infographics.
• Domain charts: Sankey diagrams in supply chains and flow analysis (immigration flows, tournament progressions, budget breakdowns); heat maps and correlation charts for relationship analysis; bell-curve distributions in performance appraisal systems; geospatial maps for logistics, healthcare outbreaks, and state-wise retail reviews.
• Chart-choice skill in practice: the bar-versus-pie and line-versus-scatter rules are the same decisions financial analysts make comparing quarterly revenue, supply-chain teams make comparing warehouse volumes, and dashboard designers make for executive review — the professional difference is reading the exact magnitude correctly, not just picking a chart.