Dashboards are a popular paradigm for data visualization in business intelligence applications. A dashboard combines a variety of base visualizations to create a single unified visual display.
However, creating an effective dashboard is a challenge since it is often difficult for analysts and designers to pick the best set of base visualizations for the task and data at hand. This can lead to unsatisfactory analytics results.
We aim to develop the Intelligent Dashboard as an interactive knowledge discovery and dashboard building system which integrates data visualization with artificial intelligence.
It employs machine learning and advanced statistics to find interesting patterns in data such as correlation, anomaly, clusters, and so on, and associates them with common data exploration tasks. such as exploration, prediction, monitoring, etc.
In the training phase analysts use a visualization recommender system to build multiple custom dashboards. This recommender system presents the set of most ‘interesting’ charts in the beginning without assuming any prior knowledge or metadata.
Over time, the system becomes smarter by learning domain knowledge using user interaction logs. In addition. it also offers better recommendations for each user based on their interaction.