Predictive analytics
Big Data & Dashboards
It all begins with an idea. This one came to live through a design thinking workshop at sovanta AG. Data Science & Forecasting have been on our clients radar — a monitoring- and forecast dashboard for their engineers was the result.
The problem
Our users were engineers who longed for an easy-to-use solution to the daring task to handle the collected data and their companies wish to step into the field of data science. We knew our users but we had to find a way to have them benefit from the big sets of data.
My role
As the lead designer I guided the client through our design process and held workshops to gather the diverse team of engineers and stakeholders. This way we all gained an understanding of the users and their pain to have more than just a pretty dashboard.
The predictive analysis tool for the system-engineers of a commodity manufacturer seeks to analyse the collected data of their produced units to forecast possible defects before they occur. A big focus was to not only show the data but to enable the engineers to analyse it in detail: filter the data as well as navigate through the various graph-axis.
As the lead designer I guided the client through our design process and held workshops to gather the diverse team of engineers and stakeholders. This way we all gained an understanding of the users and their pain to have more than just a pretty dashboard. I applied rapid prototyping methods to achieve a user-centric solution (scribbles, mockups, testings…).
Filter data sets & navigation
A Key insight was to enable the engineers to not just have a selected view of the data in shape of a simple graph but to dive deep into the data. What I see young designers doing wrong with dashboards is that they create visually appealing and simple dashboards with a selected range of pretty graphs which wouldn’t do much good to the usability.
I held workshops with focus interviews with the engineers to understand their key pain: a system which enables them to dive deep into their data and find patterns or trends. Most available CSS graph libraries have such features but we eventually decided on an easy to implement library which was able to be fed though python — allowing the data scientists on the team to prepare the data sets.
Data Science powered by sovanta AG
It was a great benefit to the project and its quality to have the scientists work on the code instead of the development team, who would have needed to be trained in the matter of data science prior to their work.
I also treasured this time working very close to the data science lab and its scientists since I quickly gained so many insights into statistics and the matter or big data. They shared their love to big Data and patterns. From an initial love for data visualization, a passion for statistics developed.
You can read more about data science at sovanta here.
Result
The scope was a small one: only a POC, proof of concept, was commissioned. We were struggling with a short timeline so we had to find solutions to get our resources focused on creating a useful product.
Success factor
Understanding the clients wish for a great example to integrate data science in their company was key.
Learning
But aside from our success factor we still had to focus on the app and producing something of value for their system engineers.