Overview
Static analysis and machine learning for automated detection of data-structure performance anti-patterns in Python.
Problem
Python programs can contain data-structure choices and access patterns whose performance costs remain difficult to identify through ordinary linting or profiling alone.
Approach
DSProfiler combines static analysis with learned signals to detect performance anti-patterns associated with Python data structures.
Results & current status
- The project is the founder's Computer Engineering B.Sc. thesis at Universitas Mercatorum.
- Coursework is complete and the thesis is in progress under the supervision of Prof. Rocco Pietrini.
Technical details
The work connects Python program analysis, data-structure behavior, performance diagnostics, and machine-learning-assisted detection.
Reproducibility
A public thesis or repository will be linked when the academic work is complete and ready for release.