This project demonstrates optimization, variable importance analysis, and contour visualization using the spotPython library.
The experiments explore 2D, 3D, and 10D analytical benchmark functions, highlighting how different input variables contribute to the output response and how contour plots can be used to interpret variable interactions.
- Contour Plots: Visualize the interaction between variables and locate maxima/minima.
- Variable Importance: Quantify which input variables have the largest impact on the output function.
- Multi-Dimensional Optimization: Apply surrogate-based optimization to benchmark functions up to 10 dimensions.
- Results Interpretation: Discuss findings from contour plots and variable importance metrics.
This project is relevant for data science, AI, and optimization research, and demonstrates practical skills in Python, surrogate modeling, and visualization.
- Contour plots showing variable interactions in 2D, 3D, and 10D functions.
- Variable importance ranking of input features.
- Multi-dimensional optimization with
spotPython. - Visual interpretation of optimization progress and results.
- Python 3
- Jupyter Notebook
numpy,matplotlib,spotpython
- In the 2D contour plots, clear variable interactions were observed (linear, non-linear, and directional).
- In the 10D function, variable x7 dominated with 100% relative importance, followed by x9.
- Several variables (x0, x2, x3, x4, x6, x8) had negligible impact, which was confirmed by contour plots.
- Optimization runs showed that focusing on the most important parameters leads to significant performance improvements.
- Surrogate-based optimization is effective for multi-dimensional problems.
- Variable importance provides actionable insights into which parameters matter most.
- Visualization (contour plots + progress plots) is essential for interpreting optimization outcomes.
Maryam Jahangir
Master’s in Automation & IT (Data Science) | Machine Learning | Optimization | AI Systems