Visualizing IoT Data With Tableau

Gert-jan.png

Author: Gert-Jan Schokkaert

VISIONWORKS MEETS JWORKS: STAIRWAYTOHEALTH

We, from VisionWorks, were asked to rebuild the visualisation dashboard JWorks used in the application they built as a result of the internal Stairway to Health project (you can find more information about that project here). We decided to use Tableau, a popular BI Visualisation tool we largely use at our clients. We developed the dashboard working around some key questions while keeping the appearance of the dashboard in line with the dashboard JWorks developed.

In the following section we will explain how the dashboard is currently set up and how to use it properly. Next, we will go over the features we can add in future releases to allow the user to go even deeper in their analysis.

DASHBOARD OVERVIEW

The dashboard is built to answer the following questions:

  1. What is the percentage of people taking the stairs or elevator at Ordina today?

  2. How is the same metric on weekly, monthly or yearly basis? What is it in absolute numbers?

  3. How does it evolve over time based on each day, week, month or year?

  4. Are people taking the stairs more this week compared to last week?

The dashboard will try to provide answers to these questions using the following three main parts. We will go over these parts and highlight which question(s) they try to answer.

PART ONE: THE TITLE

The title is what the user sees first and answers the first question. By using the colors in the title, the dashboard shows the user - in a subtle way - what the colors in the next visuals represent. There is also an option to select another day as illustrated below.

PartOne.gif

PART TWO: THE HORIZONTAL BAR COMPARISON

In part two the user can find an answer to questions two and three. The visual uses the selected day to show the division between people taking the stairs / elevator on a daily / weekly / monthly and yearly basis. When the user hovers over the chart he can also see the evolution of people taking the stairs / elevator within that day / week / month / year. Next to the chart, the total absolute number of all the observations measured is reported per period.

PartTwo.gif

PART THREE: THE MORE DETAILED AREA CHART

The third part visualises how the division stairs / elevator is evolving over time expressed in daily, weekly, monthly or yearly basis. This gives the user the possibility to look at trends and to see how the situation of today compares itself to past situations.

In the title the user has the option to change the appearance of the data (absolute or shares). The amount of periods shown (starting from the most recent period) can also be changed.

PartThreeOne.gif


When the user hovers over the chart the same horizontal bar comparison can be seen. Comparisons can be made with the period selected above.

PartThreeTwo.gif


Last feature to discuss here is how the user can change which period the chart is showing. This can be done by clicking the chart above. When you click on the day bar on the top chart the bottom chart is expressed on a day level. This also applies to the other period bases in the chart.

PartThreeThree.gif

WHAT CAN WE DO NEXT?

While this dashboard already gives an answer to the most important questions and gives the user the possibility to explore the data over time, there are still some extra things that can be developed.

The dashboard is currently built within a Tableau workbook which is using the data of the MongoDB database JWorks set up as an extract. This means we don’t have a live connection to the actual database JWorks has in their app. This brings us to the first thing we can still explore: deployment. In order to integrate the dashboard in the original application, we could publish the dashboard on the Tableau server of Ordina which is running on Microsoft Azure. Running this instance is not free so when taking a decision we should also take the user relevance in consideration: does the user really need to have a live connection to the data or does a nightly update cover the load?

Secondly we can still do a lot on the analysis part. What are the reasons why some patterns in the data exist? Do people take the stairs less when it is hot outside? JWorks recently tracked on which floor the observation is measured, allowing us to look into difference by floor. Do people take the elevator more when they need to go from floor 1 to floor 3?

We will keep you posted on further progress related to Stairway to Health. Thank you for reading and don’t forget: always take the stairs!