Our Final Result
While the machines were already generating a lot of usage data by logging every interaction, the resulting datasets were unstructured and very difficult to use to get a better understanding of customer needs and preferences. So, in order to for example determine which features are relevant to the customer and which may even be unnecessary, we had to provide our client with a data analysis tool capable of generating insights from the data they were generating.
We used Elasticsearch + Kibana to provide the client with an easy-to-use web-based dashboard for analyzing their data. Since their existing data was largely unstructured, a significant amount of preprocessing was required to find suitable structures. With the structured data our system was then able to perform simple statistical analysis and visualizations like which buttons are used how often. But for the more interesting insights we used machine learning approaches to cluster machines that are operated similarly and identify usage patterns which could for example help our client predict errors or failures of the machines, based on specific user input patterns.
- Big data
- Data mining
- Machine learning
- Data analytics
- Process mining
Generating insights about the customer’s use of printing presses for a big manufacturer. In order to provide better insights into how their customers were operating the machines, we helped our client analyze usage data of their machines. We used machine learning algorithms in order to help them understand their customers even better and to continue to make the world’s best printing presses.