Comparing multiple correspondence and principal component analyses with biomechanical signals. Example with turning...

Area of Science:

• Multivariate data analysis
• Fuzzy logic applications
• Human-machine interaction studies

Background:

• Principal Component Analysis (PCA) is a widely used dimensionality reduction technique.
• Multiple Correspondence Analysis (MCA) is less common but effective for categorical data.
• Fuzzy space windows can transform data for advanced analysis.

Purpose of the Study:

• To compare the efficacy of PCA and MCA.
• To evaluate methods for analyzing multidimensional signals from steering wheel experiments.
• To assess suitability for complex relational phenomena and large datasets.

Main Methods:

• Data transformed into membership values within fuzzy space windows.
• Application of both PCA and MCA to a 5-component multidimensional signal (steering wheel angle, hand positions, hand effort).
• Comparative analysis based on relational phenomena, computation time, and information loss.

Main Results:

• MCA demonstrates a stronger capability in visualizing complex relational patterns.
• PCA exhibits advantages in terms of analysis and computing time, particularly for large-scale data.
• Information loss during data averaging is a key consideration for both methods.

Conclusions:

• MCA is advantageous for exploring intricate relationships within data.
• PCA is more efficient for rapid analysis of extensive multidimensional datasets.
• The choice between PCA and MCA depends on specific research objectives and data characteristics.