On Thursday December 1st Tableau’s London office hosted Data Science for Internet of Things meetup with topic “Tableau – Data visualization for IoT datasets”.
IoT devices are, by their nature, data generators. They have sensors which are recording various parameters from the given domain. This data is streamed to storage and processing units. Depending on the number of deployed devices, parameters monitored and frequency of sampling, the amount of data can grow very fast. Humans can detect patterns and extract valuable information easier if data is somehow represented graphically. We can faster analyze images than series of numbers. Data visualization is therefore an important part of IoT sector.
IoT Data Visualization with Tableau
In the first part of the evening Marius Kaiser, a Product Consultant at Tableau Software, gave an insight in how Tableau can be used for data visualization and analytics.
One of the advantages of this software is that its usage does not require programming. Graphics can be defined simply by dragging and dropping certain data attributes into filter fields which get projected onto rows and columns of the diagram. This simple usage supports Descriptive analytics which gives information on data and Diagnostics and Predictive analytics which give insights in trends in data. If combined with R language, Tableau can be used for Prescriptive and Semantic Analytics which result in decision making and actions. Key areas where Tableau shows its power are Regression Analysis, Forecasting, What-If Analysis, Predictive Modeling and Clustering.
Marius demonstrated practical use of Tableau on IoT data sets, some of which were result of his own projects.
He positioned Air Quality Monitor in his flat and was observing temperature, humidity, CO2 levels and particulate matter for a set period of time. He set up IFTTT service to push each new reading to Google Spreadsheet. He then used Tableau to pull the data from the spreadsheet and create various diagrams: value change over time, location, averages etc. You can check out this project on Tableau Public.
IFTTT service is pushing Air Quality Monitor readings to Google Spreadsheet which is set as the data source for Tableau.
Peaks mark high readings of temperature and CO2. Markus explained this was caused by his overburnt pizza as sensor was positioned close to his kitchen.
Second use case Marius talked about was Engine Health Monitor. It is possible to detect e.g. jet engine failures and optimize its maintenance by having turbine sensors which are monitoring temperature and pressure and are sending samples to the cloud for analysis. Temperature degrades over time and when it reaches certain treshold that means it should be maintained. If temperature degradation continues at a faster rate, engine might be faulty and shall be replaced.
Fast temperature degradation can be sign of the faulty engine.
Similar use case is Power Plant Monitoring where analysis of turbine’s temperature and pressure results in generation of normal, warning and critical events. Tableau can display these events on diagrams grouped by location, time, warning level etc.
Another interesting example of big data analysis was detection of busiest areas of the shopping mall. Beacons are uniformly located around the mall and can pick proximity of devices and their MAC addresses. They are sending this data to the cloud which is processing it and as the result we can see how density of visitors changes in time across the mall. Tableau generated shopping mall heatmaps.
Heatmap shows areas with most frequent locations of shopping mall visitors.
Tableau for IoTDataViz
In the next part of the meetup Barend Botha, creator of IoTDataViz talked about benefits of using Tableau. Some of them he mentioned were:
- learning curve is reduced as software has intuitive UI with menu’s and controls with Drag’n’Drop
- programming is not required but can still be used (R, Python)
- supports wide range of data sources
- supports Machine Learning and AI
- allows real-time and stream processing
Heading On: Predictive maintenance for cars
Alexander Koelbl, the founder of Heading On startup presented his project about Predictive maintenance for cars. He developed a device which samples car engine temperature, RPM, battery voltage, car’s speed, location, altitude and list of devices which are draining the battery at the moment (A/C, headlights, wipers, etc…). Data is streamed to the cloud where it’s analyzed and if sub-optimal performance is detected, user gets notification via SMS.
Predictive maintenance for cars uses location, engine RPM, speed and other readings in analytics.
Industrial and nowadays home IoT installations generate huge amount of data. Representing data on graphs makes analyzing data easier and Tableau is one of the most powerful tools for data visualization, business intelligence and analytics.