M3 Master Intelligentes Bilanzkreismanagement  

Team

  • Katharina Blass
  • Robin Decker
  • Mischa Koischwitz
  • Alain Ngoufack Nguefack
  • Nguyen Do
  • The-Anh Nguyen
  • Nhu Mong Tran

Supervision

Dr. Ludwig Rauch

Data Visualization

Balancing group selection

After the user has registered for the app and logged in successfully, it is possible to choose between different balancing groups in order to only show the data the user needs to evaluate. A dropdown menu in the app’s header easily allows switching between different balancing groups.

Dashboard

The dashboard is the heart of the App, providing organized shortcuts to all features. Its layout automatically adapts to the user’s screen by displaying exactly as many widgets as legible.

Graph

As the main feature of the App, the graph visualizes various time series of the current balancing group data, such as the prediction calculated by the machine learning model. Depending on the user’s goal, it is possible to choose between three different chart types (Linechart, Barchart, Areachart). In addition, the user is able to specify the time frame and the interval of the graph. For example to display the data of each day of the last week. Within the graph edit modal, the user can decide which time series should be displayed in the graph. After setting up the graph as required, different graph configurations can be compared across multiple screens using the “open in new tab” feature. There the user sees the configured graph in full size which makes the graph even more accessible.

Explainable AI

Various factors - such as the current weekday or weather phenomena - are considerable factors for calculating the predication. Hovering over a particular section of the graphs opens an explaining tooltip showing all the essential prediction factors and their influence on the predicted value. This allows insights into the machine learning blackbox by explaining how the model came to its final results which makes the entire graph and the prediction much more comprehensible and plausible for the user.

Data Processing

Data Export

Through our user interview we figured out, that further processing of the visualized data is an important feature for the user. For example to justify decisions to superiors or stakeholders it is important to be able to quickly share the visualized information in different file formats. That’s why we implemented a second modal that allows saving the graph as a PNG or CSV file. This enables the user to share the data with others to allow insight into the data as well as to comprehend and justify different decisions.

File Management

All files created from the user’s graphs are listed in the dashboard widget as well as the files page where extra information is provided. The user can download any file for further processing, delete it or view either the file (for PNG files) or the underlying Graph (for CSV files) in a new tab.