Deciphering Reporting & Analysis Flow Through COVID-19 Dashboard

Amidst of all the craziness going around the world, I thought its better to do some research work using Power BI rather that sitting idle and washing hands.

So, in this article, I will take you through my Power BI dashboard on "Covid-19 : A Global Pandemic" and will also give you an brief idea on how to funnel the ideas in data analytics project.

This Power BI dashboard is updated on daily basis and you can find its link at the end of this article along with some other related and useful links.

The picture below is the landing page of my dashboard and as you can see, in the top left corner I have the "Date as of" and "Countries Affected" information , which gets updated on the daily basis and in the top right corner I have the some links that will take you to my YouTube channel and LinkedIn profile respectively.

Looking at the bottom left corner, we have a GitHub link that will take you to the data source from which I made this dashboard. Its an authentic data source on COVID-19 that gets updated on daily basis and then, in the bottom right corner, you have a logo of "Quantum Data" which is none other than the name of my YouTube channel.

Now, here comes the real thing :

As you can see, I have three buttons beautifully matching with the background that will take you to some specific pages.

Lets understand this.

First I have a button named "Summary", which will give you an overview of the scenario that going around the globe and obviously, you are going to have slicers that will show you data by regions/groups or, by countries and by specific date range as well.

Then , I have a button named "Deep Dive", which will take you deeper into the analysis and it could possible that you found some amazing insights by looking at it.

At the end, I have a button named "Forecast" that will show you what we can expect in the near future.

So, this is how funneling of data happens and we must always try to tell some story out of the data and this must follow a chronological order, as follows :

1. Exploratory data analysis (Exploring the data at an overall level, representing the broader picture in the form of KPIs)
2. Deep diving into the data (Looking for the sneaky factors that might be influencing the broader picture)
3. Predictive data analysis (Predicting the behavior of data in near future based on its past performance) 
4. Prescriptive data analysis (Providing valuable and impactful insights backed up by the data)

You will get a very clear understanding of the above points with a brief walk-through over my dashboard.

So, let's hop into the summary page and see what's there for us.

COVID-19 SUMMARY REPORT

As you can see in the above image, I have some big bold KPIs that shows data updated till 21st March 2020 :

1. Confirmed Cases 
2. People Recovered & Recovery % 
3. People Existing (Under test) & Existing % 
4. People died 

The KPIs also have an area chart behind them that shows the data trend and also we can see the growth from the start date (22nd Jan'2020) on the top. Upon hovering over it you can also get the "Daily Growth" information over respective cases too.

Going down the KPI band, we have two column chart stacked upon each other. The top one shows the top 5 countries by confirmed cases and the top 5 countries by death cases is shown by the bottom chart.

Then we have a beautiful ribbon chart that shows how various metrics interact with each other over time. So, from this chart we can actually visualize whether the potential curves are flattening over time or, not.

At the bottom right corner, I have a map that visually shows the number of confirmed cases by countries and as expected, the related visuals cross-filter each other and also, I have given some beautiful tooltips over the the charts that will show up once you hover upon them.

Then, at the right side of the top most ribbon, we have a "Filter" icon through which we can actually filter the whole page by specific date range and by countries/region/group and our selection will be showed at the left side of the ribbon.

The below image shows, how the dashboard looks like when we open up the filter pane and select some filters.

In case, we messed up too much with filters and just want to revert back to default page view, all we have to do is clicking the "Reset filters" button near the "Filter" icon.

To move further into the dashboard, we can either go to landing page by clicking the "Home" button in the topmost left corner and then selecting the specific report or, we can take a short-cut by clicking on the page tabs present in topmost right corner of the dashboard.

But, before moving on the the next page, let's understand the importance of colors and theme in reporting.

CHOOSING THE RIGHT THEME

A dashboard is not just about properly formatted numbers and a bunch of charts, rather it must match with the sentiments associated with the project.

This particular dashboard is about COVID-19 that took the form of a global pandemic and putted people in self-isolation and social distancing. This is indeed a dark and mysterious thing could ever happen to human race and that's the reason, I preferred a dark-reddish color background as my landing page and a dark background for other pages.

CHOOSING THE RIGHT COLOR

The color of a dashboard must be chosen carefully and the colors must be consistent throughout the dashboard to avoid confusion.

Therefore, no matter which page you go, you will find the following colors (or, their variants) for the respective metrics :

1. Confirmed Cases : Teal color
2. Recovered Cases : Green color
3. Existing Cases : Yellow color
4. Death Cases : Red color

Again, we must choose a color variant based on the dashboard background and must give the right color based on the sentiments conveyed by the metrics.

So, for example, getting recovered from this deadly disease is a positive thing and hence, I have given a green color for "Recovered Cases" and similarly, for "Death Cases", I have given a red color as it conveys loss or, negative thing, especially for this dashboard.

Now, lets move on to the second page, which gives us a deep dive analysis of the scenario :

COVID-19 DEEP-DIVE REPORT

Keeping the functionality of "Filters", "Reset Filters" and other icons/buttons constant throughout all the pages of this dashboard, we see some new chart types and some new metrics in this page :

First, we see three violin plots that shows, how the curve for different cases behaving over time along with various other metrics and the key thing that we need to monitor in this chart is "Mean" and "Median". 

So, for confirmed and death cases, as the mean and median will fall, the more good it is and for recovery, its the opposite, i.e., the more its high, the more good it is.

Then, next to the violin plot, we have an area chart that shows how the recovery growth rate is performing over the growth of mortality rate and its good if we have more peaks in this chart, but, sadly, as of now, the area under this chart is diminishing day by day.

Let's hope to see some peaks in this area chart constantly as we have seen for 12th and 21st February.

Next to the area chart, we have a pie chart that shows the split of confirmed cases, i.e., what percent of people are recovered or, dead or, under test.

It is advised not to use pie charts/donuts charts when we have more than 3 metrics because, it is visually impossible to compare them.

In the last row, we have 2 visuals side-by-side , that took population of the countries into account.

The first tree map chart visually shows the top 5 countries by percentage of population affected. So, the country highest percentage of population affected have a darker teal color and the color becomes lighter as we move to the country with least percentage of population affected.

Next to the we have a dual-axis area chart, that shows the percentage of population affected in the top 10 most populated countries and we can clearly see that how India managed to control the spread of this deadly virus as of now, despite of being the second most populated country.

Next to it, we have stream chart, that beautifully shows how the growth of three key metrics vary over time and although for a long time, the recovery growth rate was higher but, as of now, it seems to be taken over by the growth of mortality rate, which is very saddening.

Now, before we move on to our last page, lets discuss some of the important factors that we must follow while designing a dashboard :

1. Alignment of different visuals
2. Putting related visuals side-by-side
3. Utilizing the canvas space properly
4. Not showing unnecessary data labels/axes

Let's discuss the above three points one by one

ALIGNMENT OF DIFFERENT VISUALS

We must give some breathing space between two visuals and they all must be aligned and distributed properly. This approach may look very small but, it makes a huge difference when look & feel of dashboard is concerned.

Misaligned and oddly distributed visuals, misguide the viewer and they have put a lot of efforts to know the message of the dashboard.

PUTTING THE RELATED VISUALS SIDE - BY- SIDE

The visuals that are related to each other or, use same metrics in different way to deliver some related message must be putted side-by-side.

So, take an example, of the heat map chart and the dual-axis area chart in this "Deep dive" page and as both of them uses the same metrics ,i.e., percentage of population affected therefore, kept side-by-side.

UTILIZING THE CANVAS SPACE PROPERLY

As you have already understood that all the visuals in all the pages can be filtered by date range and specific country/region/group but, I don't want to waste some space in my canvas just to show some filters and that's why, I have utilized the power of bookmarks for global filtering of the dashboard.

Similarly, In the "Summary" page, I have shown the KPIs horizontally because, If I have putted them vertically then, it would have been difficult to the map visual (as it demands more width than height). But, this particular factor depends upon the nature of your visuals that you are showing in the dashboard.

NOT SHOWING UNNECESSARY DATA LABELS

Labeling your data is definitely a very good practice but, sometimes we are less focused on numbers and more focused on the trends and on those cases, adding the data labels makes the visual more clumsy and therefore, less beautiful. So, its better that we first think whether data label is a necessity for the visual or, not and then act accordingly.

For example, for the bar charts in the "Summary" pages , I have shown the data labels but, for some visuals across the dashboard, I have preferred tooltip functionality over data labels. 

Now, let's move on to the last page of the dashboard that performs predictive analysis based on the historical trends.

COVID-19 FORECAST REPORT

This page is pretty straight forward and in the first row, we have three visuals that shows the forecast data (Forecast for the next 15 days within a confidence band of 95%) along with the historical data and in the last row, we have three combo chart that shows trend line for the growth of key metrics as well as the absolute numbers.

Let's discuss about last but, not the least point that we should follow in dashboard designing,i.e., picking the right chart

PICKING THE RIGHT CHART

Other than being beautiful, a dashboard must be insightful and easy to read. So, it is advised not to use complex charts in the dashboard that are hard to read for making it beautiful.

We should always pick the right chart, i.e., if we are showing trends then we must choose the line/area chart unless there is no such other requirement.

CONCLUSION :

Although I have used Power BI platform for making this dashboard but, it really doesn't matter which platform you choose unless and until you are following the basic rules of reporting, dashboard designing and data analysis.

Let me summarize all the key steps in the form of points :

1. Get the data and make necessary transformation & massaging before you load it.
2. Perform exploratory data analysis and note down what you need to show
3. Pick the correct theme, background, color combinations & chart type
4. Show visually clear and non-complex charts
5. Do proper alignment of visuals
6. Follow the funnel, i.e, First perform "Descriptive Data Analysis" (Summary level information), then, deep diving and then "Predictive Data Analysis" (Forecasting)
7. Drawing & delivering insights (Prescriptive Data Analytics)

I hope you enjoyed your journey through the "COVID-19 : A Global Pandemic" dashboard.

Please click in the below links to connect with me over various platform :

Watch short walk-through over this dashboard : COVID-19 Dashboard Walk-through

Subscribe to my YouTube Channel :Quantum Data (Subscibe)

Link to the COVID-19 data source : GitHub/CSSEGISandData

Thank You

Udit