In this video, we will talk about visualization and its history as well as some best practices. Data visualization is not new. Although there are new tools these days that are creating new careers or new expertise in any industry including healthcare, we can see from this graph that in 1854, John Snow used this visualization to identify the wells that were causing a lot of cholera outbreaks within the City of London. You can see that there are dark bars in the middle of the graph and that indicates the frequency of cholera cases, and he was able to identify the well close to those dark bars to essentially limit the exposure to cholera and contempt and prevent the contamination of other wells. There are certain best practices in visualization and we will cover five of them in the next few slides. Before we begin, I wanted to share with you that there is really no answer to designing great visualizations, which means, it's totally flexible. It is a iterative process, especially when working with new audiences, because you have to know what the customers or the audience wants, and you have to continue to make those improvements and tweaks. And what worked in one project might not work in another project. The next five general tips will accelerate your process in achieving a great visualization. Best practice number one, audience. Ask yourself, what is your customer's end goal? And also, know how you will add value by creating these visualizations. And lastly, can a middle school student understand your work? These three questions will help you understand the objective or why you're creating such a visualization and also understand if it will actually be consumable by the audience. If you create something too complicated or something that is too clunky or difficult to navigate, the users will not end up using it regardless of how much effort you put in it. The technical teams such as the analysts or the business intelligence developers, often create sophisticated and fancy dashboards that are rarely used. And the reason for that is because the products are irrelevant or too complicated. In summary, relevant context for the intended audience and action is key. Best practice number two, techniques. You have to know the options that are available to visualize your data and apply them appropriately. In any business intelligence tool, there are options to create bar graphs, line graphs, area graphs, and scatter plots and there are many more. These are only a few examples. Think about what graphs will represent your data most effectively without making it too complicated. Use sizes, colors, shapes, and labels to represent additional dimensions in the data. What that means is instead of a traditional x and y axis, you can add in different sizes, for example, bubble graphs, have different sizes of bubbles to represent a third dimension, that could represent total revenue or total cost, and shapes could also represent different attributions such as, different genders or different types of diagnoses. Best practice number three, strategy. Develop a strategy to present and share your work. The product that you're going to develop is the most important aspect of what you have to do in visualization. However, the strategy to share that work is also quite important. And again, you have to update the visualizations for continuous use. So, think about how you would update data to refresh the application, the dashboard or whatever product you develop, because without having the ability to update the data, it's just going to become a onetime project. And lastly, be able to get feedback and make changes. So your audience needs to be continuously engaged in the product development, and you have to be able to rapidly respond to their feedback and make changes. And all these will contribute to a sustained use of your product. Best practice number four, data structure. Great visualizations depend on great data structures in the underlying data source. You have to know what fields are available and the level of granularity for those fields based on the data sources. And you also have to know how nulls or missing values are treated in the data otherwise, your visualizations are not going to be representing the right information. It's always good to have a reference to validate your findings as you're creating your visualizations if possible. Often, this is not possible because you don't know if there was any previous work done on this, and you are actually creating the first work based on a data set. Best practice number five, move on. A lot of developers spend time developing their babies which means their particular products or dashboards that were requested by the customers. And you can spend months, maybe years developing a visualization, but at the end of the day it might not be useful. And again, because this is an iterative process, it's okay, it's okay to create something that is not used as long as you learn from that experience. So, learning from your exercise and mistakes and applying your knowledge in future projects is the key point. I mean, if you want to really leverage your experience working on these projects, you could also create a community of developers and share those best practices rather than keeping it to yourself. Don't get too obsessed with your own work, you can only do it better next time. So now that we've covered five best practices in terms of visualizations, let's talk about some delivery options. Many times, the customers or the audience receive paper reports before any visualization is done. Now, those paper reports could be substituted by printouts or PDFs of the visualizations. However, it's going to be a little inefficient to create paper reports. So now, let's look at some other options. There is the email delivery option, where a user would subscribe to a routine email to get the reports automatically from your visualization or from the server that's hosting the visualization. There could also be websites, where the users could go to by clicking a hyperlink or through a portal, and then there are also posters. Many times especially in a healthcare setting, we deal with busy clinicians who don't have time to go through a website or open PDFs to look at the reports. And one of the effective uses of the data might be creating posters to let everybody know in a nursing unit or on the floor how the unit is doing, or how certain patient groups are doing. So, it's an old approach but we've seen it work quite effectively. And the key is understanding how to maximize the intended audience's engagement and action. There are many challenges and limitations when it comes to using visualizations to make improvements. Training is required for the users to understand what the visualization shows, and displaying more information leads to actually more questions, so the more data you have the more information you share, the more questions you will get, and without the ability to respond to these increase, that might actually jeopardize jeopardize the audience engagement. So, when you're creating these visualizations make sure that you are available to answer their questions, the audience's questions, and also get back to them if they have questions. Otherwise, you will lose connection to their engagement. It's also difficult to maintain data quality especially in healthcare because so much of the data is based on human input. So always look for data accuracy and ways to validate your findings.