How do you match your data to the perfect chart?

While understanding the primary type of your data is crucial, it's equally important to recognize the potential interplay between these types. For instance, if you're working with a dataset that has both temporal and spatial elements (like the sales of a product across various locations over time), the choice of visualization can drastically alter the insights gleaned. In such cases, multi-dimensional visualizations or interactive dashboards can offer a deeper, more nuanced understanding, highlighting trends that might remain hidden with more simplistic representations.

Understand the data: Clearly define the key insights we want to convey and purpose of the report Know the type of data we have like time-based, numerical, categorical, or spatial data etc Identify the relationships, trends, correlations within the data.

In the modern world, "data" typically refers to signals captured by digital (or discrete) structures, often as sequences of 0s and 1s. While splitting data types into numerical/categorical, and temporal and/or spatial is one way to think about data, another more inspiring way is to consider our natural senses. To list a few: - Hearing: processes audio signals - Sight: processes video signals - Touch: processes haptic signals - Speech: processes natural language So when assessing a data for its "type", try to think about what the information is truly about on a fundamental level.

This step is very crucial to arrive at better visualization objectives. Taking cognizance of the data type will be influential on how the data is cleaned either as text or numbers. The data we would deal mostly with can be Numerical and Categorical data. Categorical data can be ordinal eg Cold, Warm, Hot and Non ordinal eg Gender. Numerical data can be discrete which are numbers that should stand as a whole eg 2 men (cannot have 2.5 men) while continuous data are those that can allow some middle points eg height of 1.87m.

Cleaning the data and showing the correctly filtered side if it ois crucial as data changes over time and the right values ahould be looked at

It is important to understand what type of relationship we would like to display or we are looking for as per the business problem. Let us say we are looking for comparisons between different time periods, then it is good to go with bar charts. However, if we are going ahead with distribution or trend, then opting for a scatter plot or line chart respectively will make more sense. That is why it is important to see what is being solved and what is its main objective.

A chart goal is the main objective or message that you want to convey with your data visualization. It can help you narrow down the options and select the most appropriate type of chart for your data. Depending on your data and audience, your chart goal can be one of the following: ? Inform ? Compare ? Show change ? Rank ? Show relationships Choosing your chart goal is an essential step in data visualization because it helps you focus on what matters most and select the best way to present your data. By aligning your chart goal with your data and audience needs, you can create effective and engaging data visualizations that can inform, persuade, and inspire action. Like to have more information regarding bullet points.

It's important to be flexible with our goals. Sometimes, the data shows us things we didn't expect, and we should be ready to adjust accordingly.

I would reccomend to make some exploratory data visualizations during the process, trying different chart types and combinations. This way you can find out interesting relations or aspects that should be highlighted despite of the initial goal. Then narrow to the explanatory process.

Depending on the type of data at hand, we can establish relationships between them. Most people tend to lean towards Pearson or Spearman correlations, often without truly comprehending that it's just another step in the journey of Exploratory Data Analysis (EDA). I believe that the choice of visualization is linked to where and how we intend to present it. In a report or email, a chart can be fantastic, while for a customized dashboard using Streamlit, we might opt for Plotly to create something more intricate and visually appealing. The sole guideline I typically adhere to is that the information should be visually understood at a glance by someone unfamiliar with the subject matter. If this condition is met, you're on the right track.

The chart type to employ depends on data types involved and the number of unique attributes to compare. For example a pie chart is ideal to compare count of 2 or a maximum of 3 unique data attributes. A bar chart is ideal to visualize categorical and numerical data while a scatter plot is ideal to visualize two numerical data.

There are many types of charts and graphs, each with its own advantages and disadvantages. The type of chart you select depends on several factors, such as: ? The purpose of your visualization: Do you want to compare, correlate, show part-to-whole relationships, or display hierarchies? ? The type of data you have: Is it categorical, numerical, temporal, or spatial? ? The number of variables and categories you want to show: How many dimensions and levels of detail do you need to include? ? The audience of your visualization: Who will see your chart and what do they need to know? Choosing the right type of chart for your data visualization is not always easy.

When selecting the type of charts, we must consider where we intend to use them, the available space, the characteristics of the receiving screen, including potential mobile usage, and how columns are structured. Nowadays, design is paramount for effective communication, and I observe numerous dashboard errors stemming from neglecting this aspect. Once I'm clear on my communication goal, I need to determine the space required. If I'm working with numerous variables, I'd opt for a dedicated section, whereas a chart with fewer variables might occupy a third of the section. It's about fitting puzzle pieces into larger segments.

The best chart type is the one that your targeted audiences can easily understand. Consider your audience's context: whether the setting is public or private, who the key stakeholders are, their background, and the types of charts they are accustomed to. For instance, some senior business leaders favor detailed Excel tables over elaborate charts.

Selecting the right chart type for your data involves considering the characteristics of your data, the relationships you want to emphasize, and the message you want to convey. The choice of chart type can significantly impact how effectively your data is understood and interpreted. Take your time to analyze your data and consider the message you want to convey before making a decision.

It's essential to recognize that sometimes less is more in data visualization. While the temptation to use intricate designs and flashy elements can be strong, especially with advanced tools at our disposal, it's often the simplest designs that resonate most with viewers. I've often found that by periodically stepping back and removing any non-essential elements, the core message of the visualization becomes clearer, leading to better comprehension and retention by the audience.

Chart customisation is not just about the design element but can also be leveraged to highlight key trends, red flags, etc. E.g. in a bar chart you can add color conditional formatting which will highlight a specific bar in case of increase or decrease in value above a threshold. Similarly, the way you write chart titles (or add dynamic titles) and sub titles can create a huge difference in improving the messaging of your charts.

After you have created a chart, you may want to customize its design to make it more appealing and informative. There are many ways to change the appearance of your chart, such as applying a predefined style, changing the color scheme, adding or removing chart elements, and modifying the layout. Here are some tips on how to customize your chart design for data visualization: ? Apply a predefined style. ? Change the color scheme. ? Add or remove chart elements. ? Modify the layout. By customizing your chart design for data visualization, you can create more effective and attractive charts that communicate your data clearly and persuasively.

Incorporating external tools like Figma and PowerPoint to enhance your data visualizations from platforms like PowerBI and Tableau can add a significant touch of creativity and uniqueness to your charts.

In today's world, customization must adhere to the company's principles, colors, typography, and styles should align with the corporate guidelines, establishing a strong presence in our communication of importance, all while keeping the objective in focus.

When deciding on which chart to use, it is crucial to understand the appetite of your end users. If your end user is a non-technical senior stakeholder then using a very complex chart with multiple data points would not be a great idea. However, if your end user is a technical manager then experimenting with different chart types could work. The end goal should always be creating charts that are insightful, easy to understand, and that solves business problems.

I usually go with a combination chart that highlights productivity in columns and the corresponding percentages on a trend line. This combination works exceptionally well for results that need to be visualized in a progression, such as month-on-month performance tracking or comparative analysis against historical data.

Data complexity - if data is complex with multiple variables and relationships then we need to consider combining multiple charts. Visualization tools

I find it important to blend the data goal with data type steps while doing data cleaning to be able to show the right version of truth about the intended visualization. Because all 3 factors affect the type of chart visualisation and the expected outcomes that would influence the decision making process by leadership.