6.1.1.Understand data visualization - sj50179/Google-Data-Analytics-Professional-Certificate GitHub Wiki

Question

Fill in the blank: Data visualization is the _____ of data.

~~filtered view~~
graphical representation
~~final analysis~~
~~updated structure~~

Correct. Data visualization is the graphical representation of data.

Question

What are the four elements of effective data visualization?

~~Information (data), story (concept), business task (goal), and visual form (metaphor)~~
~~Information (data), topic (concept), goal (function), and visual form (story)~~
Information (data), story (concept), goal (function), and visual form (metaphor)
~~Information (data), goal (concept), business task (function), and visual form (story)~~

Correct. The four elements of effective data visualization are the information (data), the story (concept), the goal (function), and the visual form (metaphor); a successful data visualization must have all four elements.

Effective data visualizations

A data visualization, sometimes referred to as a “data viz,” allows analysts to properly interpret data. A good way to think of data visualization is that it can be the difference between utter confusion and really grasping an issue. Creating effective data visualizations is a complex task; there is a lot of advice out there, and it can be difficult to grasp it all. In this reading, you are going to learn some tips and tricks for creating effective data visualizations. First, you'll review two frameworks that are useful for thinking about how you can organize the information in your visualization. Second, you'll explore pre-attentive attributes and how they can be used to affect the way people think about your visualizations. From there, you'll do a quick review of the design principles that you should keep in mind when creating your visualization. You will end the reading by reviewing some practices that you can use to avoid creating misleading or inaccurate visualizations.

Frameworks for organizing your thoughts about visualization

Frameworks can help you organize your thoughts about data visualization and give you a useful checklist to reference. Here are two frameworks that may be useful for you as you create your own data viz:

1) The McCandless Method

You learned about the David McCandless method in the first lesson on effective data visualizations, but as a refresher, the McCandless Method lists four elements of good data visualization:

Information: the data you are working with
Story: a clear and compelling narrative or concept
Goal: a specific objective or function for the visual
Visual form: an effective use of metaphor or visual expression

Note: One useful way of approaching this framework is to notice the parts of the graphic where there is incomplete overlap between all four elements. For example, visual form without a goal, story, or data could be a sketch or even art. Data plus visual form without a goal or function is eye candy. Data with a goal but no story or visual form is boring. All four elements need to be at work to create an effective visual.

2) Kaiser Fung’s Junk Charts Trifecta Checkup

This approach is a useful set of questions that can help consumers of data visualization critique what they are consuming and determine how effective it is. The Checkup has three questions:

What is the practical question?
What does the data say?
What does the visual say?

Note: This checklist helps you think about your data viz from the perspective of your audience and decide if your visual is communicating your data effectively to them or not. In addition to these frameworks, there are some other building blocks that can help you construct your data visualizations.

Pre-attentive attributes: marks and channels

Creating effective visuals means leveraging what we know about how the brain works, and then using specific visual elements to communicate the information effectively. Pre-attentive attributes are the elements of a data visualization that people recognize automatically without conscious effort. The essential, basic building blocks that make visuals immediately understandable are called marks and channels.

Marks

Marks are basic visual objects like points, lines, and shapes. Every mark can be broken down into four qualities:

Position - Where a specific mark is in space in relation to a scale or to other marks

Size - How big, small, long, or tall a mark is

Shape - Whether a specific object is given a shape that communicates something about it

Color - What color the mark is

Channels

Channels are visual aspects or variables that represent characteristics of the data. Channels are basically marks that have been used to visualize data. Channels will vary in terms of how effective they are at communicating data based on three elements:

Accuracy - Are the channels helpful in accurately estimating the values being represented? For example, color is very accurate when communicating categorical differences, like apples and oranges. But it is much less effective when distinguishing quantitative data like 5 from 5.5.

Popout - How easy is it to distinguish certain values from others? There are many ways of drawing attention to specific parts of a visual, and many of them leverage pre-attentive attributes like line length, size, line width, shape, enclosure, hue, and intensity.

Grouping - How good is a channel at communicating groups that exist in the data? Consider the proximity, similarity, enclosure, connectedness, and continuity of the channel.

But, remember: the more you emphasize different things, the less that emphasis counts. The more you emphasize one single thing, the more that counts.

Design principles

Once you understand the pre-attentive attributes of data visualization, you can go on to design principles for creating effective visuals. These design principles are important to your work as a data analyst because they help you make sure that you are creating visualizations that communicate your data effectively to your audience. By keeping these rules in mind, you can plan and evaluate your data visualizations to decide if they are working for you and your goals. And, if they aren’t, you can adjust them!

Principle	Description
Choose the right visual	One of the first things you have to decide is which visual will be the most effective for your audience. Sometimes, a simple table is the best visualization. Other times, you need a more complex visualization to illustrate your point.
Optimize the data-ink ratio	The data-ink entails focusing on the part of the visual that is essential to understanding the point of the chart. Try to minimize non-data ink like boxes around legends or shadows to optimize the data-ink ratio.
Use orientation effectively	Make sure the written components of the visual, like the labels on a bar chart, are easy to read. You can change the orientation of your visual to make it easier to read and understand.
Color	There are a lot of important considerations when thinking about using color in your visuals. These include using color consciously and meaningfully, staying consistent throughout your visuals, being considerate of what colors mean to different people, and using inclusive color scales that make sense for everyone viewing them.
Numbers of things	Think about how many elements you include in any visual. If your visualization uses lines, try to plot five or fewer. If that isn’t possible, use color or hue to emphasize important lines. Also, when using visuals like pie charts, try to keep the number of segments to less than seven since too many elements can be distracting.

Avoiding misleading or deceptive charts

As you are considering what kind of visualization to create and how to design it, you will want to be sure that you are not creating misleading or deceptive charts. As you have been learning, data analysis provides people with insights and knowledge they can use to make decisions. So, it is important that the visualizations you create are communicating your data accurately and truthfully. Here are some common errors to avoid so that your visualizations aren’t accidentally misleading:

What to avoid	Why
Cutting off the y-axis	Changing the scale on the y-axis can make the differences between different groups in your data seem more dramatic, even if the difference is actually quite small.
Misleading use of a dual y-axis	Using a dual y-axis without clearly labeling it in your data visualization can create extremely misleading charts.
Artificially limiting the scope of the data	If you only consider the part of the data that confirms your analysis, your visualizations will be misleading because they don’t take all of the data into account.
Problematic choices in how data is binned or grouped	It is important to make sure that the way you are grouping data isn’t misleading or misrepresenting your data and disguising important trends and insights.
Using part-to-whole visuals when the totals do not sum up appropriately	If you are using a part-to-whole visual like a pie chart to explain your data, the individual parts should add up to equal 100%. If they don’t, your data visualization will be misleading.
Hiding trends in cumulative charts	Creating a cumulative chart can disguise more insightful trends by making the scale of the visualization too large to track any changes over time.
Artificially smoothing trends	Adding smooth trend lines between points in a scatter plot can make it easier to read that plot, but replacing the points with just the line can actually make it appear that the point is more connected over time than it actually was.

Finally, keep in mind that data visualization is an art form, and it takes time to develop these skills. Over your career as a data analyst, you will not only learn how to design good data visualizations, but you will also learn how to evaluate good data visualizations. Use these tips to think critically about data visualization—both as a creator and as an audience member.

The beauty of visualizing

You will find that organizing your data and communicating your results are significant parts of a data analyst’s role. In this reading, you are going to navigate different resources for effective data visualization that will allow you to choose the best model to present your data.

Inspiration is in the air

Data visualization is the graphical representation of data. But why should data analysts care about data visualization? Well your audience won’t always have the ability to interpret or understand the complex information that you relay to them so your job is to inform them of your analysis in a way that is meaningful, engaging, and easy to understand. Part of why data visualization is so effective is because people’s eyes are drawn to colors, shapes, and patterns, which makes those visual elements perfect for telling a story that goes beyond just the numbers.

Of course, one of the best ways to understand the importance of data visualization is to go through different examples of it. As a junior data analyst, you want to have several visualization options for your creative process whenever you need. Below is a list of resources that can inspire your next data-driven decisions, as well as teach you how to make your data more accessible to your audience:

The data visualization catalogue: Not sure where to start with data visualization? This catalogue features a range of different diagrams, charts, and graphs to help you find the best fit for your project. As you navigate each category, you will get a detailed description of each visualization as well as its function and a list of similar visuals.
The 25 best data visualizations: In this collection of images, explore the best examples of data that gets made into a stunning visual. Simply click on the link below each image to get an in-depth view of each project, and learn why making data visually appealing is so important.
10 data visualization blogs: Each link will lead you to a blog that is a fountain of information on everything from data storytelling to graphic data. Get your next great idea or just browse through some visual inspiration.
Information is beautiful: Founded by David McCandless, this gallery is dedicated to helping you make clearer, more informed visual decisions based on facts and data. These projects are made by students, designers, and even data analysts to help you gain insight into how they have taken their own data and turned it into visual storytelling.
Data studio gallery: Information is vital, but information presented in a digestible way is even more useful. Browse through this interactive gallery and find examples of different types of data communicated visually. You can even use the data studio tool to create your own data-driven visual.

Engage your audience

Remember: an important component of being a data analyst is the ability to communicate your findings in a way that will appeal to your audience. Data visualization has the ability to make complex (and even monotonous) information easily understood, and knowing how to utilize data visualization is a valuable skill to have. Your goal is always to help the audience have a conversation with the data so your visuals draw them into the conversation. This is especially true when you have to help your audience engage with a large amount of data, such as the flow of goods from one country to other parts of the world.

Question

Correlation charts can show relationships among data, but they don’t necessarily reveal an instance of causation. What is causation?

~~When two or more outcomes are incompatible~~
~~When two events relate to one another, but one doesn’t necessarily cause the other~~
When an action directly leads to an outcome
~~When a side effect or ramification happens after an event~~

Correct. Causation, or a cause-effect relationship, occurs when an action directly leads to an outcome.

Correlation and causation

In this reading, you will examine correlation and causation in more detail. Let’s review the definitions of these terms:

Correlation in statistics is the measure of the degree to which two variables move in relationship to each other. An example of correlation is the idea that “As the temperature goes up, ice cream sales also go up.” It is important to remember that correlation doesn’t mean that one event causes another. But, it does indicate that they have a pattern with or a relationship to each other. If one variable goes up and the other variable also goes up, it is a positive correlation. If one variable goes up and the other variable goes down, it is a negative or inverse correlation. If one variable goes up and the other variable stays about the same, there is no correlation.
Causation refers to the idea that an event leads to a specific outcome. For example, when lightning strikes, we hear the thunder (sound wave) caused by the air heating and cooling from the lightning strike. Lightning causes thunder.

Why is differentiating between correlation and causation important?

When you make conclusions from data analysis, you need to make sure that you don’t assume a causal relationship between elements of your data when there is only a correlation. When your data shows that outdoor temperature and ice cream consumption both go up at the same time, it might be tempting to conclude that hot weather causes people to eat ice cream. But, a closer examination of the data would reveal that every change in temperature doesn’t lead to a change in ice cream purchases. In addition, there might have been a sale on ice cream at the same time that the data was collected, which might not have been considered in your analysis.

Knowing the difference between correlation and causation is important when you make conclusions from your data since the stakes could be high. The next two examples illustrate the high stakes to health and human services.

Cause of disease

For example, pellagra is a disease with symptoms of dizziness, sores, vomiting, and diarrhea. In the early 1900s, people thought that the disease was caused by unsanitary living conditions. Most people who got pellagra also lived in unsanitary environments. But, a closer examination of the data showed that pellagra was the result of a lack of niacin (Vitamin B3). Unsanitary conditions were related to pellagra because most people who couldn’t afford to purchase niacin-rich foods also couldn’t afford to live in more sanitary conditions. But, dirty living conditions turned out to be a correlation only.

Distribution of aid

Here is another example. Suppose you are working for a government agency that provides food stamps. You noticed from the agency’s Google Analytics that people who qualify for food stamps are browsing the official website, but they are leaving the site without signing up for benefits. You think that the people visiting the site are leaving because they aren’t finding the information they need to sign up for food stamps. Google Analytics can help you find clues (correlations), like the same people coming back many times or how quickly people leave the page. One of those correlations might lead you to the actual cause, but you will need to collect additional data, like in a survey, to know exactly why people coming to the site aren’t signing up for food stamps. Only then can you figure out how to increase the sign-up rate.

Key takeaways

In your data analysis, remember to:

Critically analyze any correlations that you find
Examine the data’s context to determine if a causation makes sense (and can be supported by all of the data)
Understand the limitations of the tools that you use for analysis

Further information

You can explore the following article and training for more information about correlation and causation:

Correlation is not causation: This article describes the impact to a business when correlation and causation are confused.
Correlation and causation (Khan Academy lesson): This lesson describes correlation and causation along with a working example. Follow the examples of the analysis and notice if there is a positive correlation between frostbite and sledding accidents.

Dynamic visualizations

Visualizations that are interactive or change over time

The wonderful world of visualizations

As a data analyst, you will often be tasked with relaying information and data that your audience might not readily understand. Presenting your data visually is an effective way to communicate complex information and engage your stakeholders. One question to ask yourself is: “what is the best way to tell the story within my data?” This reading includes several options for you to choose from (although there are many more).

Line chart

A line chart is used to track changes over short and long periods of time. When smaller changes exist, line charts are better to use than bar graphs. Line charts can also be used to compare changes over the same period of time for more than one group.

Let’s say you want to present the graduation frequency for a particular high school between the years 2008-2012. You would input your data in a table like this:

Year	Graduation rate
2008	87
2009	89
2010	92
2011	92
2012	96

From this table, you are able to present your data in a line chart like this:

Maybe your data is more specific than above. For example, let’s say you are tasked with presenting the difference of graduation rates between male and female students. Then your chart would resemble something like this:

Column chart

Column charts use size to contrast and compare two or more values, using height or lengths to represent the specific values.

The below is example data concerning sales of vehicles over the course of 5 months:

Month	Vehicles sold
August 1, 2021	2,800
September 1, 2021	3,700
October 1, 2021	3,750
November 1, 2021	4,300
December 1, 2021	4,600

Visually, it would resemble something like this:

What would this column chart entail if we wanted to add the sales data for a competing car brand?

Heatmap

Similar to bar charts, heatmaps also use color to compare categories in a data set. They are mainly used to show relationships between two variables and use a system of color-coding to represent different values. The following heatmap plots temperature changes for each city during the hottest and coldest months of the year.

Pie chart

The pie chart is a circular graph that is divided into segments representing proportions corresponding to the quantity it represents, especially when dealing with parts of a whole.

For example, let’s say you are determining favorite movie categories among avid movie watchers. You have gathered the following data:

Movie category	Preference
Comedy	0.41
Drama	0.11
Sci-fi	0.03
Romance	0.17
Action	0.28

Visually, it would resemble something like this:

Scatter plot

Scatter plots show relationships between different variables. Scatter plots are typically used for two variables for a set of data, although additional variables can be displayed.

For example, you might want to show data of the relationship between temperature changes and ice cream sales. It would resemble something like this:

As you may notice, the higher the temperature got, the more demand there was for ice cream – so the scatter plot is great for showing the relationship between the two variables.

Distribution graph

A distribution graph displays the spread of various outcomes in a dataset.

Let’s apply this to real data. To account for its supplies, a brand new coffee shop owner wants to measure how many cups of coffee their customers consume, and they want to know if that information is dependent on the days and times of the week. That distribution graph would resemble something like this:

From this distribution graph, you may notice that the amount of coffee sales steadily increases from the beginning of the week, reaching the highest point mid-week, and then decreases towards the end of the week.

If outcomes are categorized on the x-axis by distinct numeric values (or ranges of numeric values), the distribution becomes a histogram. If data is collected from a customer rewards program, they could categorize how many customers consume between one and ten cups of coffee per week. The histogram would have ten columns representing the number of cups, and the height of the columns would indicate the number of customers drinking that many cups of coffee per week.

Reviewing each of these visual examples, where do you notice that they fit in relation to your type of data? One way to answer this is by evaluating patterns in data. Meaningful patterns can take many forms, such as:

Change: This is a trend or instance of observations that become different over time. A great way to measure change in data is through a line or column chart.
Clustering: A collection of data points with similar or different values. This is best represented through a distribution graph.
Relativity: These are observations considered in relation or in proportion to something else. You have probably seen examples of relativity data in a pie chart.
Ranking: This is a position in a scale of achievement or status. Data that requires ranking is best represented by a column chart.
Correlation: This shows a mutual relationship or connection between two or more things. A scatter plot is an excellent way to represent this type of data pattern.

Studying your data

Data analysts are tasked with collecting and interpreting data as well as displaying data in a meaningful and digestible way. Determining how to visualize your data will require studying your data’s patterns and converting it using visual cues. Feel free to practice your own charts and data in spreadsheets. Simply input your data in the spreadsheet, highlight it, then insert any chart type and view how your data can be visualized based on what you choose.

Data grows on decision trees

With so many visualization options out there for you to choose from, how do you decide what is the best way to represent your data?

A decision tree is a decision-making tool that allows you, the data analyst, to make decisions based on key questions that you can ask yourself. Each question in the visualization decision tree will help you make a decision about critical features for your visualization. Below is an example of a basic decision tree to guide you towards making a data-driven decision about which visualization is the best way to tell your story. Please note that there are many different types of decision trees that vary in complexity, and can provide more in-depth decisions.

Does your data have only one numeric variable? Histogram or Density plot
Are there multiple data sets? Line chart or Pie chart
Are you measuring changes over time? Bar chart
Do relationships between the data need to be shown? Scatter plot or Heatmap

Begin with your story

Start off by evaluating the type of data you have and go through a series of questions to determine the best visual source:

Does your data have only one numeric variable? If you have data that has one, continuous, numerical variable, then a histogram or density plot are the best methods of plotting your categorical data. Depending on your type of data, a bar chart can even be appropriate in this case. For example, if you have data pertaining to the height of a group of students, you will want to use a histogram to visualize how many students there are in each height range:

Are there multiple datasets? For cases dealing with more than one set of data, consider a line or pie chart for accurate representation of your data. A line chart will connect multiple data sets over a single, continuous line, showing how numbers have changed over time. A pie chart is good for dividing a whole into multiple categories or parts. An example of this is when you are measuring quarterly sales figures of your company. Below are examples of this data plotted on both a line and pie chart.

Are you measuring changes over time? A line chart is usually adequate for plotting trends over time. However, when the changes are larger, a bar chart is the better option. If, for example, you are measuring the number of visitors to NYC over the past 6 months, the data would look like this:

Do relationships between the data need to be shown? When you have two variables for one set of data, it is important to point out how one affects the other. Variables that pair well together are best plotted on a scatter plot. However, if there are too many data points, the relationship between variables can be obscured so a heat map can be a better representation in that case. If you are measuring the population of people across all 50 states in the United States, your data points would consist of millions so you would use a heat map. If you are simply trying to show the relationship between the number of hours spent studying and its effects on grades, your data would look like this:

Additional resources

The decision tree example used in this reading is one of many. There are multiple decision trees out there with varying levels of details that you can use to help guide your visual decisions. If you want more in-depth insight into more visual options, explore the following resources:

From data to visualization: This is an excellent analysis of a larger decision tree. With this comprehensive selection, you can search based on the kind of data you have or click on each graphic example for a definition and proper usage.
Selecting the best chart: This two-part YouTube video can help take the guesswork out of data chart selection. Depending on the type of data you are aiming to illustrate, you will be guided through when to use, when to avoid, and several examples of best practices. Part 2 of this video provides even more examples of different charts, ensuring that there is a chart for every type of data out there.

Test your knowledge on data visualizations

TOTAL POINTS 3

Question 1

Fill in the blank: Correlation charts show _____ among data.

~~changes~~
relationships
~~outcomes~~
~~causation~~

Correct. Correlation charts show relationships among data.

Question 2

When does causation occur?

When an action directly leads to an outcome
~~When an action possibly leads to an outcome~~
~~When multiple actions lead to the same outcome~~
~~When an action potentially leads to different outcomes~~

Correct. Causation occurs when an action directly leads to an outcome. Causation indicates a clear cause and effect.

Question 3

Which of the following are part of McCandless's elements of effective data visualization? Select all that apply.

~~The moral~~
~~The structure~~
The visual form
The goal

Correct. There are four elements of effective data visualization according to David McCandless. These include the information, the story, the goal, and the visual form.

6.1.1.Understand data visualization - sj50179/Google-Data-Analytics-Professional-Certificate GitHub Wiki

Question

Question

Effective data visualizations

Frameworks for organizing your thoughts about visualization

Pre-attentive attributes: marks and channels

Design principles

Avoiding misleading or deceptive charts

Further reading

The beauty of visualizing

Inspiration is in the air

Engage your audience

Question

Correlation and causation

Why is differentiating between correlation and causation important?

Key takeaways

Further information

Dynamic visualizations

The wonderful world of visualizations

Line chart

Column chart

Heatmap

Pie chart

Scatter plot

Distribution graph

Studying your data

Data grows on decision trees

Begin with your story

Additional resources

Test your knowledge on data visualizations

Question 1

Question 2

Question 3