Dendrogram and Principal Component Analysis Applied to Geotechnical CBR Data to Remove Data Noise

Dendrogram analysis involves creating a hierarchical tree structure, which visually represents the relationships between different data points based on their similarities or differences. By clustering similar data points together, dendrogram analysis can help geotechnical engineers gain insights into the underlying relationships present in the data.Principal Component Analysis (PCA) identifies patterns that encode the highest variance in the data. PCA involves aggregating information inherent in multi-dimensional data, representing it with a reduced number of new variables. Dendrogram and PCA analysis application in geotechnical engineering areshown using California Bearing Ratio (CBR) test data. The relationship and groupings within the many data attributes not typically apparent are shown. Engineers typically uses only the CBR test value but the inter-relationships between the various "pieces" contributing to the CBR result is not evident. The PCA quantifies the key component of the tests where most variance occurs. A correlation matrix is used to show all analysis methods point to similar conclusionand industry practice of (incorrectly) using density ratio as the key parameter in quality control during earthworks construction.

Keywords: Dendrograms, Hierarchical cluster analysis, CBR tests, Density ratio, Principal component.