Multicollinearity analysis is a popular statistical technique, deals with the situation where multiple variables are linearly correlated. The presence of a strong correlation can be easily detected by this technique. Correlation matrix, route-square, tolerance, and variance inflation factor are major outputs of multicollinearity algorithms. The present study is concentrated on the Southwestern hills (Ajodhya, Garpanchakot, Biharinath, and Susunia) of West Bengal. Seventy-seven soil samples are collected from the top of the hill to the foothill pediment area according to the change of elevation and slope. The location of sample points has been determined by the use of a GPS handset. Elevation of the sample points has been measured by use of altimeter and further rechecked by overlay operation on SRTM digital elevation model in GIS software. The slope of the sample locations has been determined by clinometer compass. Soil and geomorphology are deeply correlated to each other. Two sets of data with common geomorphological and different soil variables have been employed for the study. Soil variables have been determined in the pedological laboratory through different soil sample analysis methods. This is the pioneer's attempt to find out multiple correlations among the soil and geomorphological variables as they are closely related to each other. This attempt is very unique and appropriate because strong multiple linear correlations have been detected in the variables of sand, silt, clay, organic carbon, and organic matter through route square, tolerance, and inflation factor values. Multicollinearity must be resolved in different ways, but the principal component analysis is one of them.
Mots clés : Techniques|variables|correlation|soil|component
A104850AA