🏫 This project is part of a Masters thesis in the Visual Analytics and Visualization (VAV) lab is part of the Faculty of Computer Science, Dalhousie University, Canada.
📙 View on DalSpace library
📹 Video demo of a use case
- Finding delay-related patterns affecting turnaround performance
- Reducing the cognitive load of exploration of the mined patterns by means of visual analytics
This project was originally developed using a log of ground handling operations at at Halifax Stanfield International Airport collected by Assaia Apron AI.
For showing that the proposed framework can be used in other domains as well, we also used a dataset of chronic diseases. This dataset is collected by the Flaredown application and is publicly available at Kaggle.
- Using data mining (sequential rules and frequent itemsets)
- Visualization of patterns with a novel matrix-based approach
🌱 The main contribution of this work is a novel approach for visualizing sequential rules w.r.t. to the partial order of antecedents and consequents.
- Domain expert feedback on visual prototype
- User test
-
Two sets of analytical tasks
- Similar questions
- Different tools
- text vs. visual prototype
-
NASA-TLX for measuring workload of each set of tasks
-
Are difference in averages statistically significant?
- Paired t-test
- (alternative) hypothesis: the true mean difference between the paired samples is not equal to zero
- Paired t-test
-
Analytical tasks to assess if the users can successfully use all features
-
SeRViz is a novel visual analytics tool for mining and exploring frequent patterns. Based on our experiments it reduces the cognitive load of users for the said tasks compared to the popular off-the-shelf data mining tool, SPMF.
👩🏫 Tutorial SeRViz Mining, filtering and overview
👩🏫 Tutorial SeRViz Sequential rule matrix
👩🏫 Tutorial SeRViz distribution analysis and breakdown
👩🏫 Tutorial SeRViz frequent itemsets
cd server
pip install -r requirements.txt
python main.py
👉 Since this project uses SPMF library, you need have Java installed. An instance of SPMF v2.42c is located at server/pattern_mining/mining/thirdparty/spmf.jar (uploaded since it is required for live demo on Heroku).
We developed a low-fidelity prototype in early stages of visual design using dash. The dashboard is located at server/mining/quick_dashboard.py.
This research was enabled in part by support provided by and DeepSense 🌊 and Halifax Stanfield International Airport:airplane:.
Asal Jalilvand. SeRViz: an Interactive Visualization Framework for the Analysis of Sequential Rules and Frequent Itemsets. Master's thesis, Dalhousie University, Halifax, NS, Canada, 2021.