Let's talk about the world of data visualization. This may raise the following questions: What is visualizing? What are visualizations used for? What types of charts fit which data sets? What tools allow us to visualize? Keep reading and resolve all your doubts about this area.

To visualize is to transmit something to our receiver. If we take it to the spiritual side, it could be transmitting love, happiness, anger, and/or frustration; from the statistics side, it can represent an increase or decrease in the global economy; from the color side, red could represent alert or danger, while green can mean success. We receive all this information daily without asking ourselves how that result was reached. Why is red associated with danger and green with success? Now focusing on the technological side, visualizations serve to show a graphical representation of data. Here, large data sets or datasets are taken and organized in a way that can offer relevant information to the receiver. For example: we have a dataset from a school, where all the students' grades with their respective subjects are recorded. Thanks to visualization, these data could be taken, entered into a table, and visualize which class has the worst average, or which subject students have the best grades in. It could also show the top 10 students with the best grades in the entire school. Taking a bigger example, we have a dataset with all the schools in Chile and want to know which are the best schools by region, or which are the 5 schools with the lowest grades in the Los Lagos Region. All that and much more can be achieved in a few minutes thanks to data visualization.

Types of charts

A statistical datum is a visual representation of a series of data. Today there are many types of charts, which are divided into qualitative, referring to qualities that cannot be numerically represented; here we find ordinal and categorical types. On the other hand, we have quantitative, which refer to quantities or numerical values. Among them, we have discrete values that only take integer numbers, such as the number of students or number of schools, and continuous values that take any value within a range, for example, an average age or an average grade. Having this difference clear, we can show the most important charts used to create visualizations. (Reference article here)

Bar chart‍

It is a graphical representation on a Cartesian axis of the frequencies of a qualitative or discrete variable. It can be found in both vertical and horizontal formats. There are three types of bar charts: the first is a simple bar chart, the second is a bar chart composed of different fields, and the third is a stacked bar chart where all fields are in the same stack.

‍

Histogram‍

It is a bar chart used to represent the frequencies of a continuous quantitative variable, identified by having its bars joined together. 

Line chart‍

It is a graphical representation on a Cartesian axis of the relationship between two variables, clearly reflecting changes produced over a period. They are usually used to represent temporal trends. 

‍

Pareto chart

It is a type of vertical bar chart ordered by frequencies in descending order that identifies and prioritizes data.

Pie charts‍

Also known as pie or cake charts, it is a circular representation of the relative frequencies of a qualitative or discrete variable that allows, in a simple and quick way, comparison. This chart cannot have fewer than two fields or more than five fields.

Pictogram‍

It is a chart that represents the frequencies of a qualitative or discrete variable through figures or symbols.

Scatter plot‍

Here the relationship between two variables is shown on a Cartesian axis.

Cartogram‍

It is a map where statistical data by regions are presented either by placing numbers or coloring different areas according to the data they represent. 

Creating a visualization

To start a visualization, you must immerse yourself in the context and nature of the data you will work with, as this will give you clarity on the scope of the aspects that can be done in the tool you are working with. This point is very important as it will align expectations with users and avoid errors in the final product. The prior study of the data accounts for 70% of the total time required for a visualization project. If the data comes with errors, the entire process will be flawed. 

‍

‍

As a team, we must develop a proposal based on the best practices of data visualization. This will be our Mockup, which will provide a visual reference to the Product Owner and users of the layout of elements on the screen and the representation of the KPIs previously defined in the different charts.

We recommend arriving with at least 2 Mockup proposals to iterate together with the receiving team, making it easier to reach a final agreement, which will translate into the list of requirements and will be the basis for the progress checklist.

Once the mockup is defined, the Storyboard must be worked on. This work must be based on the user group and their needs regarding the dashboard. The same visualization can be used by both a manager and a department head; however, their interests and the granularity of the metrics to observe for each profile are completely different. A manager expects to observe the development of metrics at a macro level, while a department head will monitor the evolution shown on the dashboard day by day. This simple fact implies that the filters, elements, and actions within the dashboard will differ. Given this, a storyboard must be created for each user profile group that will use the dashboard. This section will also be reflected in the requirements document for the progress checklist.

There are different tools to create visualizations. Amazon Web Services offers a tool called QuickSight, which is a fast, cloud-based, fully managed business intelligence service. QuickSight allows creating and publishing interactive visualizations. Its payment method is pay-as-you-go, allowing connection with a wide range of datasets such as MySQL, PostgreSQL, MariaDB, Aurora, SQL Server, Redshift, among others. It also supports Excel files, files stored in Amazon S3, GitHub, Twitter, among others. (For more information click Here).

Another visualization tool is Power BI. It consists of a set of business analysis tools that originated as an integrated solution within Office 365. It allows connection to hundreds of data sources, simplified data preparation, and ad hoc analysis generation. The advantages of using this tool are its flexibility, as it allows extracting information, organizing it, transforming it, and combining data from multiple sources. It creates interactive and customizable visualizations. Additionally, it is multiplatform. (For more information click Here ).

Finally, we will talk about Tableau, which offers more effective, secure, and flexible comprehensive data analysis. It is a business intelligence platform that transforms your data to drive actions based on information. Tableau was designed for the individual but adapts to any company. (For more information click Here).

There are 4 ways to work in Tableau:

● Tableau Desktop: Considered a “model of excellence” in visual analysis, this is where analysis is performed. Thanks to its easy-to-use interface, Tableau Desktop revolutionized the business intelligence sector.

● Tableau Online: It is the answer for cloud self-service analysis, requiring no server management. It is secure, scalable, and requires no hardware maintenance.

● Tableau Prep: It is a tool that prepares your data for analysis and allows more people to combine, clean, and shape data quickly and confidently.

● Tableau Server: It is a business analysis tool ideal for business-related environments, allowing sharing and managing data and information on-premises or in the public cloud.

Visualization Development 

You own a bookstore in your city and are generating positive profits, so much so that you are thinking about expanding the store, which implies bringing in more books, but you don't know which books are the best sellers, the best-selling genre, nor which books sell the least. For this reason, you decide to store the data in a dataset and then connect that data to a tool that can answer all your questions about the books. 

The first thing a visualization tool will do is identify whether the fields correspond to measures or dimensions. Measures correspond to quantitative values and are usually represented in green, while dimensions are qualitative values and are usually represented in blue. Additionally, the application contains a list of all possible charts that can be used; depending on the user's need, the appropriate chart will be applied. In the case of the bookstore owner, he wants to identify which books are the best sellers and which books are the most purchased. 

Once the data type and chart to use are identified, the user will have the task of identifying what functions the tool has stored. One of its most used functions is the filtering system, where you can filter by date (year, month, or day), filter by genre (female or male), filter by geographic data (continent, country, or city), etc. You can also create new fields called “calculated fields.” These arise from calculations the user needs to create the visualization without modifying the source dataset. We can calculate the total of two original dataset fields, calculate the average of those two fields, concatenate two string fields into one string field, create calculated fields indicating the difference between two dates, etc. 

‍

‍

There are also functions that indicate the maximum or minimum value in a dataset. For example, this is useful to calculate the total sum of each book genre sold and observe that adventure novels report the highest sales while fairy tale novels report the lowest sales. Additionally, these tools have functions that show a range of values depending on a value entered by the user. For example, if the user wants to know which sales are greater than $35,000, he enters that value in the application, and it will return all values greater than that or, conversely, all values less than that. You can also configure the application to display the top x that the user requires. Using these functions, the bookstore owner can see a top 10 of the best-selling books in less than 5 minutes and identify the genre of each book.

Assuming the dataset also contains customer information, with that data, we can determine who buys books, see which gender (male or female) makes more purchases, create age ranges (child, teenager, young adult, adult, or senior), and identify in which age range the highest purchase indices are found. This will help the bookstore owner identify which books to buy, since if the child index is low, it is likely not profitable to buy many children's books. 

On the statistical functions side, these tools do not fall behind. Most have integrated statistical functions such as summation, average, difference between data, converting data to percentages, difference between percentages, standard deviation, variance, percentages, among many others. For example, Tableau allows analyzing trend lines showing the R-squared value and the p-value, as shown in the following illustration. 

Visualizations impact users thanks to the great amount of interaction they can perform with their data, and color is an essential element when delivering a product. For example, as indicated at the beginning, we could make a chart that separates book sales into three sections. The first section would be poor sales, assigned the color red; then normal sales, which have little impact, assigned the color yellow; and finally, the sales that matter to the bookstore owner: high sales, assigned the color green. So every time the owner updates the chart, he will look at which books appear in the green section—those are the ones he will order from the book factory—while those in the yellow section he will wait for to sell what is on display, and those in the red section he might apply discounts to sell them. He could also rearrange and place the unsold books at the front so that when the customer enters, the first thing they see is those books, possibly generating a future sale. Colors provide a visual impact to the user, but they should not be overused. It is recommended not to use more than 6 colors, or the user will lose focus on the information.

The last step is to create the dashboard, where in a single view, the charts created in the previous steps are inserted. Depending on the application used, actions allowed in the dashboard will vary. Some applications allow creating filters, for example, a filter by year. So if the user selects the year 2019, all charts change their data to that year. If the user changes to 2020, the user will see data corresponding to that year, and so on. You can also choose which charts change and which do not, as sometimes you don't want all charts to change. When creating the action, you select which chart(s) will follow the filter.

Let's imagine we have created a cartogram showing all the cities of Chile. If the user wants to see how many books were sold in a particular city, such as Viña del Mar, an action is created in that chart, and that action applies to the other charts. So if the user selects any city, all charts will change their data related to that city. Depending on the application, some allow uploading images and URLs to complement the information delivered in the dashboard. The dashboard creation process ends here. Now the bookstore owner is clear about the decisions he will make for his business, knows which types of books to buy, which books he might put on sale, and also knows which cities in Chile generate the most income and which cities he should consider closing. 
‍
Ready to transform your data into strategic decisions?

At Kranio, we help you implement data visualization solutions tailored to your business needs, allowing you to interpret information clearly and effectively. Contact us and discover how we can boost your data strategy.