Comment créer des visualisations de données à la fois axées sur les données et centrées sur l’humain ?

Understand the domain of your audience To ensure you use your visuals to tell the right story, you need to understand the domain of the data (Sales, Finance, and Health) you will be presenting to your audience. This understanding helps you during your presentation by ensuring flow in conversation.

Before creating any visualization, it's crucial to understand your target audience. Consider their ✔level of expertise, ✔background knowledge, and ✔expectations. Key questions to consider: ✔What are their goals and needs? ✔What kind of information are they hoping to gain from the data? ✔In which context will they view the visualization?

One thing that really helps is figuring out who you're talking to. ✔Understand if they know a lot about the data or not and what they're hoping to get from it. This helps you design visuals that match what your audience needs and can understand. ✔When designing, keep things simple but informative. ✔Use clear labels, don't make it too busy, and pick visuals that make the info easy to get. ✔Add interactive stuff, like pop-up info or filters, so people can explore on their own.

The harsh truth is that most of the visuals will not get viewed heavily unless you answer the question of what business people want to know. ✔ Business people may want to know clear actionable KPIs, how many customers we have in the last quarter that used certain licenses. This may help them execute item now. ✔ Make dashboard with short-term actionable dashboard that business can use immediately. ✔ To create dashboard that is useful for long term goal, you need to first prove yourself that you can build something that stakeholders want to see. → If you don't listen to your audience, your dashboard will be frozen in the freezer.

In the data visualization process, it is crucial to tailor the representations according to the target audience. Segmenting users into groups such as Executive, Analyst, Operational, and General can assist in creating targeted dashboards. Executives prefer direct metrics, analysts require exploratory details, operational collaborators seek specific information, and the general public needs a more narrative and explanatory approach. Adapting the data for each group is crucial to ensure that the visualizations are accessible, understandable, and relevant to their specific audience, making the information more impactful and useful.

Having the right dataset is the first step in data analysis and data visualization. If the dataset chosen is wrong, analysis and data visualizations will be wrong these will there for lead to wrong decisions by the decision makers.

Using the right data is always important but if you are under the influence of bias even that won't help much In one of my past personal projects, I was creating a dashboard on COVID influence all over the world. The dashboard was based on the official Covid dashboard and soon I realised that the dashboard always gives the feeling as India was one of the worst impacted countries in the world since the analysis was based on the basis of total covid cases which was not correct. Instead, the analysis should have been based on the population density, no. of Covid cases / 100000 people. I set my home page as wrt population density with an option to check with total population as well, even the correct data can be manipulated so beware

Also note: Choosing the right data does not imply cherry picking only the range that conforms to your theory. Anomalies can be just as informative as the rest of the data.

Clean and reliable data are the top considerations for generating anything useful, but as data scientist, we need to be responsible for generating visualizations that rely on solid statistics. A beautiful visualizaton is just like a “sketch” or a study. The right data allows you to view meaningful findings and a great source for more advanced statistical analysis

Esse é um grande desafio não coletar propriamente os dados , mas saber acolher quais usar, principalmente em Um Contexto de abundância de informações

No particular visual is perfect for all scenarios although some visuals are more frequently used than others, however, the data analyst should ensure the right visual is used to communicate the message to the passed across. An example will be the use of a gauge to display target vs actual.

Select the most effective visual representation for your data by considering the nature of your information. For trends over time, a line chart might be suitable, while comparisons could benefit from bar charts. Choose visuals that enhance understanding; for complex relationships, consider scatter plots or network diagrams. Keep it simple and aligned with your audience's preferences. Test different visuals and iterate based on user feedback to ensure the chosen visualizations convey your data accurately and are easily interpretable.

I believe that using the best visuals for relevant KPIs enhances the dashboard adoption rates. I strongly believe that before making the actual dashboard, one should definitely do the dashboard wireframing.

Using the best visual to represent the underlying data is the most crucial part of a Data Analytics Process. The visual should be simple, self-explanatory, and human-centered. On the other hand, a data-driven approach is essential to remove anomalies or any other kind of unwanted information from the data. Hence, a data-driven approach followed by a human-centered approach would be ideal in order to select the right visual for the data.

We can consider the below questions to use the best visuals. 1.What is the type, scale, and distribution of your data? 2.What specific message, story, or insight do you want to convey? 3.How can you use visual hierarchy to emphasize important elements? 4.How can you apply contrast, alignment, and consistency in your visual design? 5.How can you prevent distortion, clutter, or unnecessary noise in your visualization? 6.Have you selected the most suitable visual technique for your data? 7.How can color and size be used effectively to enhance visual impact? 8.Does the visual design align with principles of user-centered design? 9.Are visual elements accessible to diverse audiences, including those with visual impairments?

One thing I have found helpful is to have multiple brainstorming sessions with target audiences of the dashboards. This helps to set right context and narrative for your visualization.

A captivating narrative elucidates the apparent story conveyed by visuals. Audiences seek to understand your findings, desiring a clear grasp of the concept behind the vibrant display, and readable text facilitates a seamless comprehension process.

Narratives can have positive or negative influences on your audience. It's important to use narratives carefully and in a way that doesn't misrepresent the data or disenfranchise your audience. Moral narratives, for example, are the most powerful but also the easiest to mislead the audience with. Think visualizations on gun violence, renewable energy, health care costs, etc.

The narrative I create when speaking to audiences or writing stories is there to inform and engage perspective (different from perception). I agree with eliciting curiosity, emotion, and action but I prefer to accomplish this without persuasive tactics. The information should be rendered as not to be manipulative --solely informative.

When choosing data for a human-centered, data-driven visualization, consider context and narrative. Instead of just presenting raw numbers, provide background information to help users understand the significance of the data. For instance, if showcasing sales figures, include context about market trends or product launches. Craft a narrative that guides users through the data, emphasizing key points and insights. This not only enhances comprehension but also makes the information more memorable and impactful, fostering a deeper connection between the data and the audience.

Feedback from the user is great! It helps us curate the dashboard/viz specifically for the target audience. It helps trim the insignificant items.

Nothing can kill a dashboard more than poor performance. If there's too much stuff on a tab, eliminate objects to ensure things load faster.

Select a group of data champions in your organization that can help you to test what you have created. That way you can get feedback from actual users and determine if the visualization is actually answering questions

By considering the below questions we can easily test and refine our data visualizations. 1.Have you actively sought feedback from your target audience, peers, or experts? 2.How do you measure the overall effectiveness of your data visualizations? 3.How accurate is the information presented in your data visualizations? 4.What methods are you using to test the appeal and engagement of your visualizations? 5.Have you conducted usability testing to assess the ease of interaction and understanding? 6.What specific metrics are you using to evaluate the performance of your visualizations? 7.How are you measuring the outcomes or impact of your data visualizations? 8.How will you address the visualizations based on feedback and evaluation results?

At this stage, it is important to talk to people who have a neutral view of your data. If you work with the data for a long time, you sometimes lose sight of the big picture. That's why it's also important to get input at this stage. Check the data again for validation. In this phase, you also ask yourself typical questions that once again focus on the essentials and check whether the data is also valid for the stakeholders.

Being upfront and transparent about the limitations and timeliness of your data visualization is essential. If the graph only incorporates data or respondent opinions from dense urban centers then that information should be readily apparent instead of forcing viewers to dig deeper into a bit of the fine print to discern the timing.

Sharing your visuals on different platforms like LinkedIn provides the opportunity for people (other professionals) to provide feedback and share their perspectives on your visualization. This feedback is an opportunity for you to learn, it also creates an opportunity for others to learn through your post.

I find that many that are purely in the data visualisation side of things typically do not foster relationships with the media. It makes sense to have a media arm within your organisation to show journalists and relevant websites that you have uncovered in your data visualisation and infographics.

In my experience, it is always good to distribute our visualizations through a range of channels, including social media, blogs, and presentations, to enhance visibility and accommodate diverse audience preferences. Actively monitor audience reactions and feedback to refine and improve your visualizations over time.

Be transparent with your process taking raw data to a contextualized or visualized data output. Document your data acquisition, processing, exclusions, and decision points throughout that lifecycle. Consider the question: Are the steps I took clearly laid out so that others can replicate my process to get a repeatable result?

To create data visualizations that are both data-driven and human-centered, I believe one should focus on clarity, simplicity, and context. Ensure the visual accurately represents the data, but also prioritize usability by making it intuitive and accessible. Tailor the design to your audience’s needs and use storytelling to guide them through the insights, allowing them to easily interpret and engage with the information.

Be customer-centric so you understand what users are trying to get out of a visualization. Make sure data is accurate and real-time. Choose simple visuals when the purpose is informing and be really creative when the purpose is celebrating or engaging.

Sometimes, even if we have created the most beautiful data visualization and dashboard, they are of little use to senior management. In such cases, do not stick to specific features, instead analyze the big picture and create a visualization for the business insights which may fit into new requirements. Change is constant, and this is true even for Data visualization. What worked for me last year, is not relevant anymore today.

There is a saying that "if it takes more than 10 seconds to explain a dashboard, you've done it wrong". Visualisations should be easy to interpret. Senior managers should look at your dashboard and know within seconds what the story is and what action is needed. Simplicity is key.

Crafting data visualizations with a human touch involves understanding your audience deeply. Start by defining clear objectives—what story do you want the data to tell? Keep it simple, choose visuals wisely, and add interactive features for a personalized experience. Think of it as storytelling; structure your visualization to guide your audience through a narrative. Accessibility matters, so ensure everyone can engage. Iterate based on feedback and strike a balance between aesthetics and functionality. It's about creating a visual journey that not only informs but captivates.