Tableau Maps, Actions, and Chart Types

8.1 Module 2 Wrap-Up, the Midterm Syllabus, and Quiz-Doubt Clarifications

8.1.1 Module 2 Wrap-Up: Where the Course Stands

This session closes module two of the course. Module two was the conceptual Tableau module — how a Tableau start works, and how the building blocks of a Tableau workbook are put together. It covered the product family, connections, data types, the visual interface, dimensions and measures, hierarchies, sorting, grouping, parameters, combined fields, sets, titles and captions, exporting, metadata management, and filtering — in other words, everything needed to build an individual worksheet properly.

Today covers two things from the official flow: how we create maps in Tableau, and how we do some ad hoc analysis. On top of that readout, the session adds a few extra concepts — the actions that make sheets talk to each other, several chart types (scatter plots, word maps, bubble charts, funnel charts, and waterfalls), and a revision of the design principles (pre-attentive attributes, semiotics, and storytelling). The reasoning is practical: the more of these you learn now, the easier dashboarding will be later, because dashboards are nothing but sheets plus interactivity plus design.

The regular quiz is skipped this time — the class has just written one — and the time is reused to clarify the doubts students raised about it. The order of the session is deliberate: doubts first (they are the freshest in everyone's mind), then a quick recap, then the new content.

The plan for the next few sessions is announced up front, so the class knows what this session is building toward. Module three starts from the next session and runs over classes nine, ten, and eleven: dashboard building. Everything learned so far — individual sheets, how to build different graphs — comes together there, and those three classes are a mix of a little theory and mostly practice. The closing notes of the session add more detail: the three dashboarding classes also include stories, and after Tableau the course moves to Python libraries in modules four and five — matplotlib first, then seaborn and bokeh — because visualization is not done only by drag-and-drop tools.

8.1.2 The Midterm Syllabus and Exam Format

A student opens the session by asking about the midterm syllabus, and the answer sets expectations for the whole exam season.

Why does "case based and conceptual" change how you should study? Because it moves the emphasis from remembering menu paths to reasoning about a situation. The skill being tested is the same one a working analyst uses: given a business question and a dataset, which visualization approach helps, and in what order do you apply it. Reading the first eight chapters for their ideas — chart choices, design principles, what each analysis type is for — pays off more than memorizing where buttons live in Tableau.

8.1.3 Exploratory versus Explanatory Analysis

The first quiz doubt is the difference between exploratory analysis and explanatory analysis — the two big modes of working with data that the course introduced in its very first session.

Exploratory analysis (exploring the data, written EDA for short) is the phase where we explore things: what can happen here, what happens if we try something else, which variable relates to which. The goal is discovery — you do not yet know what you are looking for, so you try directions, look at many charts, and let patterns surface. Explanatory analysis (explaining the data) is the phase where everything is already done: the analysis is complete, the graphs are ready, and the visualization gives absolute clarity, which helps us draw a conclusion. The relationship stands out — "this is going up, this is going down" — and that is the result. The definition comes from the very first session: explanatory analysis has the goal to confirm or refute the initial hypothesis, explain the underlying relationship, and draw conclusion from the data.

The analogy used when the concept was first taught is the deciding picture: when you go tracking (exploring terrain), you are exploring — you walk around, check the ground, try paths. When you come up with a map at the end, you have explored everything, you give a very clear answer, and you take a decision. That is explanatory: the map is the conclusion, and the decision rests on it.

The connection to design comes back later in this session. During exploratory work you should mostly avoid heavy visual styling — too many colors and highlights can mislead you while you are still looking for patterns. During explanatory work, on the other hand, you want the pre-attentive attributes (color, size, position) to make the story visually clear, because there is a specific message to deliver. The mode of analysis decides how much design is appropriate.

8.1.4 Data Cleaning versus Data Scrubbing

The second doubt is the difference between data cleaning and data scrubbing — two everyday terms that sound interchangeable but are not.

Data cleaning (the broader term) takes care of everything: correcting errors, correcting inconsistencies, handling missing values — and it makes sure the data is absolutely ready for all our aspects, probably including data visualization. Think of it as the full preparation pass that leaves the data trustworthy for every downstream step. Data scrubbing is only a tiny subset of that: it just removes the unwanted data. Scrubbing takes out what you do not want — an unwanted table, an unwanted row — but it does not ensure accuracy and consistency.

The everyday picture: cleaning is the full spring-clean that makes a house ready to receive guests — floors, windows, broken furniture, everything. Scrubbing is one small task inside that, like taking out the bin of things you no longer want. Removing the bin makes the house neater, but it does not fix the leaky tap. The analogy breaks where the scope is different: scrubbing is not a "wrong" or useless practice — it is simply a smaller practice, and the exam question is about which one matches the bigger job.

8.1.5 Data Inaccuracy versus Data Inconsistency

Related to the same quiz discussion is the difference between data inaccuracy and data inconsistency — and which of the two is the most common, and most fundamental, challenge.

Data inaccuracy is when something is fundamentally wrong: you typed 20 instead of 200, or the gender field says female instead of male. The whole report, the whole concept goes for a toss — we are presenting a wrong thing to our stakeholders. Data inconsistency is when the data is there, but different sources disagree: one table says pin code 560777, another table says pin code 560778 for the same place. Inconsistency is a real and very common problem, but it sits much later in the lifecycle.

Why does accuracy come first? Think in order of operations. If the data itself is wrong, there is no point worrying about inconsistency first — you are working on wrong data. Data analysis cannot even pass the next stage if the data is wrong, so accuracy is the most basic, most fundamental thing when we start our visualization. If a data analyst is asked what to fix first — a wrong value, or two tables that disagree — the top priority is the wrong data. Both problems are common and often go side by side, but accuracy comes first.

The mental picture: a wrong number is a wrong foundation — whatever you build on it (reports, charts, decisions) is untrustworthy. Inconsistency is two blueprints that disagree about a window: annoying, and worth resolving, but pointless to argue about until the foundation itself is correct.

8.1.6 Heat Maps and Hierarchical Data

The next doubt is about heat maps: are they used for visualizing data with a hierarchical structure, and do they also help with grid-based data?

A heat map is used when we want to uncover hidden relationships — that is its purpose. Each cell of the map carries a value, and the color of the cell (the "heat") carries the magnitude, so your eye can scan a large matrix and spot patterns instantly: which combinations are high, which are low, where the structure lives. The standard textbook treatment agrees — the heat map (also called a matrix chart) is built for rapid pattern matching across a matrix of categorical combinations, using a color scheme where stronger saturation or darker shade ranks the values.

When the data has levels — level one, level two — the hierarchical heat map lets you zoom into further levels, and that is where the real benefit comes: you can drill, you can zoom in, you can move from the big picture to the detail without changing the chart type. A grid, by contrast, is just a plain table — a structure where you can have text. For grid-based data a heat map will not serve the purpose; grid is just like any simple table. The heat map's strength is mainly for hierarchical data, where it helps with the relationships.

8.1.7 How Multiple-Choice Questions Are Designed

The session also explains why quiz questions feel ambiguous — a frustration the class voices almost every week. MCQ options are never at the extreme ends: they will never be so absolutely black and white that the answer is pointable at first glance. That is why we always struggle: we zero it down to two options and then have to pick between them. Every option is made to look somewhat correct; the art of the question is in the scope and the qualifiers that make one of them the better fit.

The instructor's advice to the class: in these sessions, ask more questions rather than assuming — asking in class may help more than assuming the answer in an exam, because a question asked now is a concept understood before the exam. It is acknowledged that if the next quiz happens, the questions may be made a little more direct, but the MCQ style itself will keep looking like every option is correct — that is how MCQs work. The instructor also firmly encourages this kind of questioning: it is how the class learns.

8.2 Recap: What Module 2 Built

8.2.1 The Foundations: Connections, Data, and the Interface

A quick recap orients everyone before the new content, and the missing pieces are covered by the earlier material. The point of the recap is to fix the mental model: a Tableau workbook is built from the data upward — connect the data, understand its fields, then place fields on shelves to build sheets. Everything new in this session (maps, actions, extra chart types) sits on top of those foundations.

The Tableau product family was scoped earlier — including the products that exist and how Tableau is installed — and the first step of any work is the data connection: how we usually connect our data before starting visualization. The connection step includes the two modes:

• Live connections, where the figures refresh on the fly — every time the underlying database changes, the view updates with it.
• Extracts, where Tableau takes a one-time copy of the data and analysis runs against the local copy — faster at build time, but the copy is static until you refresh it.

The six data types in Tableau were covered (text/string, number, date, date and time, Boolean, and geographic values), and the data interpreter — the helper that activates when the data is not in a clean table format, for example when a spreadsheet has title rows or stray text above the real table. We saw how to view the data before visualization (the data source page preview), and how the Tableau visual interface looks: what the grid (the columns and rows shelves) is, what columns and rows are, and where the fields usually come — from the data pane on the left.

The two big classifications are the heart of the model:

We also learned about hierarchies (and why we create them — to drill from a level like country down to state and then city without rebuilding the view), sorting, and grouping (combining values that belong together into one category); how to pass parameters into reporting (a user-controlled input that flows into filters, calculations, and titles); and how to combine fields (joining two or more fields into a single dimension so sorting and grouping can work across them). These went beyond the handout, but they help in building, and some are used very frequently.

Real-world: there are many ways to get Tableau for practice. There is Tableau Public — the free, public-facing version that is perfect for learning — and there is also a one-year free license for students. It genuinely helps to download a version and try things yourself, because watching someone else drag and drop a few fields here and there will not teach it; the try-it-yourself part is the whole point.

8.2.2 Export Formats and the Individual-Sheet Toolbox

We also covered how to export a worksheet — turning a finished sheet into a shareable artifact. The formats covered earlier include PDF and image; the session's recap quiz remembers GIF as another one — a reminder that export options are broader than the two most obvious ones and are worth checking in the menu when a stakeholder asks for a deliverable.

The recent material also included metadata filtering (hiding and unhiding fields in the data source so the view stays clean), drill-down hierarchies (navigating through the levels of a hierarchy at runtime), and calculations (calculated fields — new measures defined by formulas over existing fields) — all at the individual worksheet level.

This session moves up a step on the ladder that leads to dashboards: maps first, at the sheet level, then actions, which will matter when sheets come together into dashboards — because an action is exactly the glue that makes one sheet affect another, and that only becomes visible once you have more than one sheet to connect.

8.3 Maps: Geo-Spatial Fields, Filled Maps, Shape Maps, Layers, and Options

8.3.1 Geo-Spatial Fields, the Globe Icon, and Geographical Roles

Tableau is intelligent enough to recognize some fields as geo-spatial fields — fields whose values are places: states, countries, cities, pin codes, and so on. The tell is the icon: every field that Tableau expects to be geographical carries a small globe icon in front of its name. Anything with a globe is a geo-spatial field, and Tableau can directly show it as a map. This is the single fastest fact-check in the course: no globe icon, no out-of-the-box map.

The reason this works without any setup is that Tableau has its own database underneath: it has captured longitude and latitude values for millions of locations and knows that California, New Mexico, and all the other states are states. With that database, no work is needed from us: double-click a state field and Tableau automatically builds the map — the geography, the borders, and the projection are all supplied by Tableau's built-in data.

Alongside the built-in database come geographical roles. When the field is placed, Tableau maps it to a role — for example, the field named "state" is assigned the state geographical role, because the field name strongly suggests it. The roles matter because they decide how the field is drawn: a field treated as a country draws country borders, a field treated as a city draws points, and so on.

The globe icon follows the role: the moment a field gets a geographical role, it gets the globe, and the map options unlock. If the icon is missing, check the role before checking anything else.

8.3.2 Filled Maps

A filled map (the field map) fills the area of the graph with a chosen value — the classic choropleth-style map where whole regions become the marks. With a state-wise map in place, we drag a measure such as profit onto the sheet and every state gets filled with a value. A measure is the thing with the numerical value — from the earlier material, something we can aggregate — and the map assigns each state's fill color according to its profit.

The filled map is the fastest way to see the geography of a number across the whole country at once. It works best when the question is about regions as areas: which states, which districts, which countries are high or low. The encoding is simple — area position plus color intensity — which is exactly why executives can read it without training.

8.3.3 Shape Maps

The second way to show values on a map is the shape map: instead of filling the areas, we place a shape at each location. We can choose the shape — the demo uses a star (an asterisk), and the plus sign is another option — and we can make the size bigger, so the size can carry meaning.

Which option you pick depends on your requirements and what your business stakeholders need: either fill the field, or show the shapes. Filled maps are better for continuous area coverage and reading the overall geography of a value; shape maps are better when you want point-level emphasis, when locations are scattered, or when a single map must show several measures side by side through different shapes.

8.3.4 Map Layers

Map layers are extra content we can add to or remove from a map — the background context that sits underneath your data. Right-click the map, choose map layers, and the options appear. The demo walks through the available layers:

• Coastline — zoom in and the coastline shows up, giving the geography a real edge.
• Terrain — the physical relief of the land (hills, valleys) appears in the map.
• Streets, highways, and routes — the road network, shown when the analysis is about movement or place-level detail.

Some of the fields are US specific, but many layers can be shown in the same maps. There are also different map styles: light, normal, satellite (a real satellite image — well, not literally real, but a satellite-style image), outdoor, and street; colors can be changed, and there is a dark option for presentations on dark backgrounds.

The right combination depends on the use case: pick and choose what you want to include or exclude in your map, and the change is a click of a button. Layers are decoration in service of the message — they never change the data, only the context around it.

8.3.5 Map Options

Map options are the tools and controls around the map — the navigation and selection furniture. You can:

• Zoom in and out — move from the whole country to a city block.
• Pan (move the graph around) — slide the viewport across the map without changing the data.
• Pin or unpin the view — lock the current position and zoom so it stays fixed when the data changes.
• Select — mark areas of the map with a rectangle, a circle, or a free-flowing shape, and the selected marks stay selected or can be unselected.

The map options themselves are editable: open the options and decide what you want and what you do not want — for example, removing the search control from the map when the end user does not need it. Everything is about customizing the look and feel, based on your preference or your client's preference.

Real-world: map styling matters in client work. A logistics dashboard might need the street-level layers when zoomed into a city, while an executive view over states reads better on a clean filled map with few layers — the layers and styles exist so the same data can be dressed for the audience. One dataset, two audiences, two maps: the map options and layers are how you switch between them without touching the data.

8.4 Bringing Your Own Data: Roles, Unknowns, and Location Fixes

8.4.1 Bringing Your Own Data: The India Cities Example

The demo starts from a hand-made dummy Excel file with India cities and some kind of sales values. Opening it in Tableau shows the same fields — cities and sales — but the map options are disabled. The reason is the field type: Tableau reads cities as a string field (the ABC icon), and strings mean text — Tableau says, look, this is a text field, I do not know how to make a graph from this. The number (the hash icon) is obvious, the text is not.

The fix is the role change from the previous section: we tell Tableau to start treating cities as a city. This is the manual override of the geographical role — the field name "city" or "cities" may be absent or different in the Excel file, so Tableau cannot guess, and we assign the role ourselves.

The edit locations dialog shows the problem for each unknown:

• Bangalore is not recognized; the name in Tableau's database is Bangaluru, so we remap Bangalore to Bangaluru, and after the fix Bangalore appears.
• Delhi looks correctly spelled, but Tableau's database stores the city under a different name — the recording of the demo garbles the exact stored name, but the point is the same: the spelling you typed is not the spelling the database holds, and the remap fixes it. Open edit locations, select the unknown cell, and tell Tableau which database entry "Delhi" should be treated as; after the remap the city appears.
• Hyderabad has a spelling problem — one letter is extra or missing (the demo data was missing an "a") — and fixing the spelling through edit locations places the city.

Fix each unknown one by one through edit locations, and every value starts appearing. That is how we bring our own fields onto the map: change the role, double-click, and repair the names Tableau cannot match.

The instructor's encouragement after the demo: if you spend two hours on Tableau yourself, you will become a champion — it is that user friendly. The last few sessions may feel like a lot of information, but you do not have to remember everything on day one; the point is to know what Tableau is capable of. The tool is learnable by doing, and the two-hour self-practice threshold is the honest price of mastery.

8.4.2 The Country Fix: US City Data

The second demo uses US city data, and it shows a second class of problem: the country is wrong. The field must be treated as city again, and double-clicking brings up unknowns. Opening edit locations shows why: because the previous graph in the session was India, Tableau is assuming this new data is also India — so the first thing to fix is the country, which we set to United States.

The country check is the cheapest fix in the whole workflow, which is why it comes first in the debugging order below. It is also the easiest to overlook, because the error message never says "wrong country" — it just says "unknown."

8.4.3 Ambiguous Names and the Longitude–Latitude Fix

The remaining problem is an ambiguous name: Boston is a repetitive name in Tableau's database — Boston probably exists in other countries too, or is a duplicate — so the name alone cannot be resolved. Tableau sees the word "Boston" and cannot tell which of several Bostons you mean. One fix is giving the right name (a fuller name that is unique, or the country-qualified name); another option is pinning the location by longitude and latitude — the coordinates remove all ambiguity because they identify one point on the planet.

So even when a name match fails entirely, the coordinates rescue the location. This is the last resort of the workflow, but it is a guaranteed one: any place on Earth has a longitude and latitude, and entering them pins it.

8.4.4 The Debugging Workflow for Unknowns

The whole debugging workflow for map unknowns, in order:

Fix all of these and your graph will start appearing. The order is not random — each step is cheaper than the next, and each fixes the most common cause first: country assumptions are free to check, roles take one click, names take a remap, coordinates take a web search. Working top-down solves most unknown-location problems in step two or three.

So the summary of maps: if the data is correct, map it to the right role; then fix the errors, either by name or by longitude–latitude; and this is how maps work in Tableau.

8.5 Actions: Making Sheets Interactive

8.5.1 Why Actions Exist

Actions are the feature that makes sheets interactive with each other, and they are the bridge from individual sheets to dashboards: these will be very useful when we create dashboards, because the dashboard becomes interactive — you click on one menu and the sheet below gets updated or reflected. There are three kinds covered today: filter actions, highlight actions, and URL actions.

Each is created from the worksheet menu (worksheet, then actions) when working on a single sheet, and from the dashboard menu when working on a dashboard. The action dialog asks for the same things every time:

8.5.2 Filter Actions

A filter action filters the target sheet based on what you select on the source sheet — the target's data actually changes: values that do not match the selection disappear. Typically you have two sheets — they can be any sheets. For example, one sheet shows India states and another shows cities; you can say: if I click on a state, show me the cities of that state.

To set it up: come to worksheet, come to actions, and define the action. The dialog asks for your source sheet — where the action actually runs — and your target sheet. In the demo, the source sheet is sheet one and the target shows the product subcategory, with the setting "show all values based on the selection I do." The action can be based on any pair of sheets: sheet three and four, or any others — you tell Tableau, based on which sheet, what action is to be done on the target sheet.

8.5.3 Highlight Actions

A highlight action does not remove data — it draws attention to the matching values and grays out the rest. The demo builds a product category-wise sales chart, colored by region (the region is added as the color shade), and the action's source sheet is this product category chart with the second sheet as the target. With the trigger set to select, clicking a region on the source sheet highlights only the matching slice on the target: clicking the east region of furniture highlights exactly the east furniture bar, and everything else grays out; clicking central technology leaves only central technology colored.

The difference from the filter action is visible: with a filter, the previous example's values were going away entirely; with highlight, you still see the grayed-out values, but the selection lights up — for example office supplies in west.

The trigger is configurable. The demo changes the highlight from select to hover, and the target updates as the mouse moves over the source sheet. Hover works for highlighting; for URL actions (below) the same choice is a warning rather than a virtue. The dialog also offers broader settings — all sheets, all fields, all values — for the scope of the action, and existing actions can be edited at any time (edit highlight, edit filter).

When to pick which: if the question is "show me just this slice," use a filter; if the question is "show me where this slice sits inside the whole," use a highlight.

8.5.4 Actions at the Dashboard Level

The same actions are created at the dashboard level, because a dashboard is a collection of all the sheets. When you open a dashboard, it shows you all the sheets that exist — with two sheets in the workbook, it automatically shows two; had there been five, it would have shown five. You drag and drop both sheets onto the dashboard, then use dashboard, then actions to create the action there.

The demo creates a third highlight: the source is the first sheet, the target is the sheet below, and on select the outcome shows in the target. Clicking on the first sheet, the sheet below updates automatically — clicking technology lights up only the technology slice (the yellow one) in the target.

Imagine a dashboard where the user clicks one thing and the other sheets get impacted automatically: that is how actions work, and that is why they are very, very useful for dashboarding.

8.5.5 URL Actions

A URL action opens a browser page when the user triggers it. Setup: worksheet, then actions, then add action, go to URL. The dialog asks what you want to happen on this worksheet.

The demo builds the classic example: a state map where clicking a state opens the Wikipedia page for that state. The first attempt hardcodes the URL — the California Wikipedia page is pasted in — and the result is the trap: clicking Nevada still opens California, because the action was hardcoded to one page. The fix is to remove the hardcoding and tell Tableau to take the state from whatever the user clicks: the state field value gets embedded into the URL as a parameter. After that, clicking Texas opens the Texas page, clicking California opens California — whatever you click, the page follows.

The trigger options matter for URL actions. On select, every click jumps to the page — which can be too much: you accidentally click and you are taken to Wikipedia. The menu trigger fixes that: clicking a state now shows a hyperlink menu, and the option is on the user — they decide whether to follow the link, and clicking the hyperlink opens the page. There is also hover, which the instructor explicitly warns against for URLs: hovering on any state would open a new page — "probably is too heavy for the system." The demo also shows the browser behavior option: open the URL in a new tab or a new browser page.

Wikipedia is only an example. If your business has web pages for the products you showcase, you can give the URLs for that particular value, and the user can go to the website and come back; you can point to product pages, FAQ pages, or pages carrying the latest statistics — there are n number of use cases. The URL can be customized per value through the embedded field parameter, so the same action serves every member of the field.

8.5.6 Recap: Sheets First, Actions on Top

The workflow order is stressed: first create your sheets — first your visualization, your KPI should be clear — and only then add actions; actions sit on top of the sheets. Then decide the behavior: a filter (on select or on menu), a highlight when there are multiple colorings, or a URL action, choosing the source sheet, the destination sheet, the value, and how the web page behavior should work. When you are doing an action on a worksheet, come to worksheet; when you are doing an action on a dashboard, come to dashboard, drag and drop your sheets, and define it at the dashboard level.

Real-world: the URL action is the pattern behind drill-through reporting — a sales map where clicking a territory opens the territory's detailed web report or product catalog, so the dashboard stays clean while the detail lives in the browser. Filter and highlight actions, in the same way, are what make delivered dashboards self-service: the user clicks one menu and the sheets below update — the modern expectation for organizational dashboards.

8.6 Tooltips, Encoding Choices, and the Show Me Panel

8.6.1 Tooltips: Deciding What the Reader Sees

A tooltip is the little popup that appears when the pointer rests on a mark, and it is fully customizable. Everything shown in it is a choice: in the demo the tooltip shows the state, and we can decide exactly what we want there. Drag any field into the tooltip — sales, profit, or both — and it appears when the user hovers. The demo adds profit, then sales, to the same tooltip.

This is the pattern for every encoding decision: decide what you want to show, then drag and drop — any coloring you want, drag the field into color; anything on size, give it the size; labels work the same way (the map labels the states, and a discount value can join the label too). The instructor's framing: sit down, decide what you want to show, then drag and drop, and it becomes very, very easy.

Real-world: in a delivered dashboard the tooltip is where the supporting numbers live without cluttering the chart — the map stays readable while the hover carries state, sales, and profit detail. The tooltip is the fine print that never has to be on the poster.

8.6.2 Color, Size, and Label Choices

The same drag-drop logic applies to color, size, and labels. In the map demo, shading moves from sales to profit by dragging profit into color; sales and profit can both go into the label together by dragging both there; color and size do not have to encode the same field — one variable can drive color while another drives size, when you want to show two variables at once.

The rule of thumb: one channel, one job. Use color for the categorical split (regions, categories), size for the magnitude, labels for the exact numbers the audience needs, and tooltips for everything else. When two variables must appear together, give them two separate channels — color and size — rather than stacking both on one.

8.6.3 The Show Me Panel and Its 24 Templates

The Show Me panel at the top of the window is a three-by-eight grid — 24 chart templates in total. Every chart you build is based on one of these templates, and with innovation we can create many other kinds of visualization on top of them, but the fundamental stays the same: ultimately we use one of these templates.

Depending on the fields chosen, some of the 24 get enabled and others stay disabled — Tableau switches off the templates that cannot represent your current field selection. The maps earlier appeared in this panel too, enabled precisely because the field was geo-spatial: the moment the field carried the globe icon, the map templates lit up.

The 24 templates also explain why the chart-type demos later in the session are so quick: scatter plots, bubble charts, and the rest are not being built from scratch — they are templates being selected with the right fields in place.

8.7 More Chart Types: Scatter Plots, Word Maps, Bubble Charts, Funnel Charts, and Waterfalls

8.7.1 Scatter Plots and Drop Lines

A scatter plot is straightforward: pick two measures and plot one against the other — one measure on the x-axis, one on the y-axis, and each row of data becomes a point at its pair of values. The textbook treatment calls it the chart for revealing correlation, clustering, and outliers between two quantitative variables: where the points crowd into a band, the two measures move together; where a point sits far from the crowd, it is an outlier worth investigating.

The demo builds a sales-versus-profit scatter, broken down by subcategory, to see the relationship between the two numbers. The extra feature is the drop line: a reference line from a selected point down to the axes. When you click a point, the drop line shows what value of x and what value of y that point has — useful because reading those values with the naked eye is not always easy.

The drop lines are editable: show them only when a point is selected, or show them always — the demo notes that always-on can be too much, against what we learned in data visualization — and you can choose whether the drop line appears on the x-axis, the y-axis, or both.

8.7.2 Word Maps: Text Size and Color

A word map (word cloud style view) looks complicated but is not. The mechanism: drag the field whose values you want as words into the text shelf. The demo puts state into text, so all the state names appear. Then choose the encodings: color by sales and size by sales — the moment sales goes into size, the words scale with the value — and select the text mark type from the Show Me panel (by default Tableau uses whatever automatic view it prefers, so choosing text from Show Me is the step that makes the word map).

Color can encode one thing while size encodes another; just choose wisely so it does not confuse the reader.

8.7.3 Bubble Charts

A bubble chart shows values as circles whose size carries the number. The textbook view: the bubble plot extends the scatter plot by encoding more variables on the same mark — position still places the value, area carries magnitude, and color carries a category. Select two fields — the demo double-clicks state and sales — and from the enabled chart types pick the bubble chart; then color it by state so the view is a state-wise bubble chart. The bigger the circle, the bigger the sales, because the numeric measure drives the bubble size.

When circles are close together and hard to tell apart, add the sales value as a label: in the region-wise version, the labels make it easy to read that west is slightly better than east.

As with the word map, color and size can encode different variables — decide what you will choose color for and what you will choose the size of the bubble for.

8.7.4 Funnel Charts

A funnel chart is also easy to make. The demo builds a region-wise sales funnel: choose the region field and the value field, put the size on sales, color by region, and choose entire view. The funnel shows each region's contribution sized by its sales value — the classic narrowing shape reads as a pipeline from the largest contributor at the top to the smallest at the bottom.

The instructor admits live that one step is missing from the demonstration — "sometimes you miss a very basic fundamental step and then it goes for a toss" — but the recipe is: choose what field you want, choose the entire view, and then choose on what you want the funnel chart to be made.

8.7.5 Waterfall Charts via the Gantt Mark

The waterfall chart uses the Gantt mark: you choose the Gantt chart, show the value, and it makes the waterfall for you — again very straightforward. The waterfall shows how a starting value accumulates or steps down across categories, which is the classic chart for profit build-up or cost break-downs. The textbook describes it as pulling apart a stacked bar: a starting point, a sequence of increases and decreases, and a resulting ending point — each bar floats so you see the net effect of each step.

In Tableau, the trick is the mark: choose the Gantt chart, drop the value in, and the floating bars line up into the waterfall automatically — no layout math needed.

Real-world: these chart types map to common business questions — scatter plots for relationship checks between two metrics (is profit moving with sales?), bubble charts for share-of-total comparisons across regions, funnels for pipeline or contribution views (which stage loses the most prospects?), and waterfalls for showing how a total builds up step by step (headcount, cost build-up, profit build-up).

8.8 Design Principles for Dashboards: Pre-Attentive Attributes, Semiotics, and Storytelling

8.8.1 Pre-Attentive Attributes: Color, Size, and Position

That is what pre-attentive attributes do — we can use color, we can use size, we can use position to immediately draw attention. "Pre-attentive" means the brain registers the difference before any deliberate attention: a red bar among gray bars is seen before it is "read." These were learned in the earlier sessions: how to make sure the user looks at what we want, how to use hierarchy — where to place what, for example the top-left corner is usually the title.

The standard list is richer than the three the instructor names — the classic catalogue includes motion, spatial position, intensity, hue (color), line width, line length, shape, orientation, enclosure, added marks, curvature, and size — but the course's three (color, size, position) are the ones that do most of the work in dashboards. The point of all of them is the same: they create a visual hierarchy that tells the reader what to see first, second, and third, without a single instruction being printed.

8.8.2 The Brain Processes Visuals Quickly

The reason pre-attentive attributes work is that our brains are wired to process visuals quickly. The demonstration: a set of items where three were counted without any forcing — the moment three items were made gray, the mind could count them within seconds, without instruction. The textbook version of the same experiment is the "count the 3s" exercise: in a grid of digits, counting the 3s takes real effort until the 3s are highlighted — then the answer is visible at a glance. That is the proof that the visual channel is fast, so we have to use these tools wisely and let them help our visuals.

Color, shapes, and position play an important role; we have to learn to avoid too many colors and choose shapes with care. This is not natural to all of us, but it is a skill we will continue to sharpen, and it helps in designing charts and making them easy.

There is also a mode-of-analysis angle, tying back to the quiz concepts: during exploratory analysis you should mostly avoid heavy pre-attentive styling, because bright highlights can steer your own reading before the patterns are known; during explanatory work, the story is fixed, and pre-attentive attributes exist to make that story visually clear.

8.8.3 Semiotics: The Study of Signs and Symbols

When we deal with signs and symbols, the whole thing is called semiotics — the study of signs and symbols. The point: use signs and symbols wisely when we present our data, so the reader reads the intended meaning without effort.

A semiotic mistake is a mismatch between the sign and the meaning readers attach to it: using a red-green scale where the audience expects a stop-go meaning, using icons that suggest one thing while encoding another, or choosing symbols that need a legend to decode. The discipline is the reverse: pick signs whose everyday meaning already matches the message, so the chart explains itself.

8.8.4 Storytelling as a Narrative

Storytelling has to be narrative based: highlight what we want our user to look at, and make it interactive. The building blocks we have been building are exactly what makes it interactive — sheets, filters, and actions are the mechanisms a story is told through.

The narrative arc in the demo: start with our sales five years back, then say what we did, then show a trend — a time series showing how profit went up, went down, and this was the reason — and where we are now. That is how you build a story, so your audience can engage. The structure is the classic one from the storytelling sessions: a clear beginning (where we were), a middle (what happened, with the reason), and an end (where we are now) — the same beginning-middle-end shape that makes stories stick in memory, applied to data.

8.8.5 Data Exploration and Discovery

Data exploration and discovery follows a workflow in any visualization tool, not just Tableau. First identify what questions you want to navigate for; then use your filters and drill down — give relevant filters for users to navigate; then see the patterns in that. Questions first, navigation second, patterns last.

The order matters because it inverts the beginner instinct. Beginners open the tool and start clicking, hoping patterns appear. The workflow says: the questions come before the filters, and the filters come before the patterns — a filter is only useful if you already know what you are looking for, and a pattern is only meaningful if it answers a question you actually asked.

8.8.6 Plan with the Stakeholder Before Opening Tableau

The strongest piece of professional advice in the session: do not sit in front of the machine and randomly drag and drop fields — with so many fields you will randomly drag and drop and get lost. Go to the machine at the last. First spend time on the board with your stakeholder: what exactly are they looking for; identify what you will present; agree the KPIs; find out what filters they will like — do they want a filter on date, on quarter, on the zone. Identify these things first, then build. After the build, let them play with your dashboards. The ideal order is: do the homework, discuss with the stakeholder, then come to Tableau at the end.

Real-world: this stakeholder-first workflow is how reporting engagements run in industry — the analyst walks in with agreed KPIs and filter requirements, builds once, and hands the users a self-service dashboard instead of a slide deck that misses the question. The board time before the build is what separates a dashboard people use from a dashboard people glance at.

8.9 The Road Ahead: Dashboards, Stories, and Python Libraries

8.9.1 Module 3: Dashboards and Stories in Classes 9–11

Module three starts with the next session. The next class will bring a little theory — the do's and don'ts of dashboards, what kinds of dashboards exist, and what mistakes we usually make — so it is semi-theory and semi-practical. The two classes after that are purely hands-on dashboard building. Over the next three classes we will finish the Tableau part, which includes dashboards as well as stories. There is no class during the exam week; the class meets again a week after that and goes into dashboarding.

Everything from this session is the preparation for that module: the sheets are the material, the actions are the interactivity, and the design principles are the standards the dashboards will be judged against. When the dashboarding classes start, the building vocabulary is already in place.

8.9.2 Modules 4 and 5: Python Libraries

After Tableau, modules four and five give a quick overview of what the Python libraries offer — because visualization is not done only by these tools. Module four is on matplotlib, and classes 14, 15, and 16 move to seaborn and bokeh. (The recording garbles the second pair of names, but the course plan is clear: matplotlib first, then the seaborn and bokeh libraries.) The course continues with those libraries for a couple of sessions and then continues the journey.

The roadmap matters beyond scheduling: it completes the course's promise that visualization is a skill, not a tool. The same design principles taught around Tableau — choosing the right chart, encoding with intent, telling a story — carry over to Python, where every chart is built by code instead of drag-and-drop.

8.9.3 Closing Q&A: What Lands in the Exams

The session closes with the exam question that students care most about: will the practical topics covered today appear in the exams? The answer draws a clean line between the two exams.

The first exam is case and application based — you will be asked how you would apply these concepts to a scenario; it will not ask you to give a Tableau command and say what you would click. So maps, actions, and the other practical things will not be there in the first exam. The second exam, the closed-book semester exam, is further away and its format is not finalized, but the instructor is definite that these Tableau practical things will not be there either.

Real-world: this mirrors how tool skills are assessed in industry — you are judged on choosing and defending the right visualization for a business question, not on recalling menu paths. The interview question is never "which menu opens map layers"; it is "your regional head wants to see profit by state — what do you build and why."

Exam Guidance Summary

• The midterm syllabus is the first eight chapters. Questions are case based and conceptual: the paper gives a scenario and asks what you would consider for it and how you would approach data visualization — expect questions about how you would proceed, not about tool clicks.
• The first exam (EC1) will not contain practical Tableau operations: no "give a command in Tableau and do an action," no what-would-you-click questions. It tests application of concepts to scenarios.
• The second exam (EC2), the closed-book semester exam, is still far away and its format is not final, but practical Tableau details are expected to stay out of it as well.
• Quiz concepts to be sure of: exploratory analysis explores while explanatory analysis confirms or refutes the initial hypothesis and draws conclusions from the data; data cleaning is the broad practice covering errors, inconsistencies, and missing values while data scrubbing only removes unwanted data; data inaccuracy (fundamentally wrong values, like 20 typed instead of 200) is the most fundamental starting point and outranks data inconsistency (same place recorded as pin code 560777 in one table and 560778 in another); heat maps uncover hidden relationships in hierarchical data and are not meant for plain grids.
• MCQ design: options are deliberately not black and white — you will often narrow it to two plausible choices; expect every option to look somewhat correct. If anything is unclear, ask in class rather than assuming.
• Maps concept to be ready for: geo-spatial fields carry the globe icon, Tableau has a built-in geo database, roles can be assigned manually, and unknown locations are fixed through edit locations (check country, check field type, check spelling, or supply longitude and latitude).
• Actions concept to be ready for: filter actions remove values on the target sheet, highlight actions gray out non-matches, URL actions open parameterized web pages — and sheets must be built before actions are added on top.
• The three dashboarding classes (9, 10, 11) include stories, and the next class is semi-theory: dashboard do's and don'ts, dashboard types, and common mistakes. There is no class in exam week; the course resumes a week after with dashboarding.

Key Industry Applications

• Real-world: Tableau Public and the one-year free license are the standard ways students and practitioners get hands-on Tableau practice; the try-it-yourself experience is the whole point.
• Real-world: Tableau's built-in geo database means state and city maps build themselves from clean data — the field map (filled map) and shape map are the everyday choices for territory dashboards.
• Real-world: map layers and styles let the same geography serve different audiences — street-level detail for city operations, clean filled maps for executive summaries.
• Real-world: the own-data workflow is the norm for departmental requests — a hand-made Excel with city names and sales arrives, the role is fixed, and unknown or misspelled locations (Bangalore to Bangaluru, Hyderabad, ambiguous Boston) are repaired via edit locations or longitude–latitude lookups from Google.
• Real-world: URL actions are drill-through links — clicking a state can open a product page, an FAQ page, or a page with the latest statistics, with the value embedded in the URL as a parameter.
• Real-world: filter and highlight actions make dashboards self-service: the user clicks one menu and the sheets below update, the modern expectation for organizational dashboards.
• Real-world: the stakeholder-first workflow — agree KPIs and filters (date, quarter, zone) on the board before touching Tableau, then let users play with the dashboard — is how professional reporting projects are run.
• Real-world: after Tableau, the same visualization work moves to Python — matplotlib, seaborn, and bokeh — so the course covers both tool-based and code-based visualization.