Modeling of soil shear strength using multiple linear regression (MLR) at Penang, Malaysia
Bala Balarabe, Andy Anderson Bery
Abstract
This paper presents the multiple linear regression (MLR) soil strength models developed from electrical resistivity and seismic refraction tomography. The multiple linear regression technique is used to estimate the value of dependent variables of soil strength based on the value of two independent variables, namely resistivity and velocity values. These parameters were regressed using regression statistics technique for generating multiple linear regression model. The analysed model results of MLR model which is based on estimation of model dependent parameters (resistivity and velocity) calculated for P-value at significance level of 0.05 is 0.01572 and 0.01163, for soil’s cohesion parameter and 0.01966 and 0.02534, for soil’s friction angle parameter. Two MLR model equations were developed from the statistical analysis. The forecast accuracy of the MLR model is conducted for verification on the second stage. Based on these statistical analysis results, a new soil’s strength model from geophysical data set for near surface study were developed. The soil’s strength models developed using MLR is reliable to image the subsurface in two-dimensional form, which covered more region compared to traditional method.