Probe drilling prediction on geological condition using logistic regression of R
Loading...
Date
2016-06
Authors
Muhammad Fitri Mohd Akhir
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
In this research, probe drilling data and JH rock classification system are chosen to study the prediction of geological data and analyze the data by using statistical analysis. The parameters that involved in this case study are geological texture (sand, gravel and clay), groundwater flowrate, drilling speed, and joint. All of these parameters are chosen based on previous study and criteria in JH rock classification. Rock mass classes of JH system are categorical data type and have been used as dependent variable. Rock mass classes also have been changed from alphabet to numeric for ease the statistical analysis and declared as RC. In this statistical analysis, multinomial logistic regression has been applied to find the prior variables that influence to RC in both NATMs. Multinomial logistic regression is applied when there is a single categorical dependent variable and more than one independent variables. For validation of the model analysis, model-fit test was tested by using deviance and -2 log 2 to find the best fit model of parameters that influence to RC. The lower the -2 log 2, the better the model. In this case study, model joint 2 has been chosen as the best fit model of parameters that influence to RC in both NATMs. The coefficients in the model joint 2 in both NATMs are tested by using Z-test to check the significant influence of parameters toward each all of RC. The parameters that influence to all RC are taken as the coefficient in the formulation formula of RC in both NATMs. Both formulas are used to predict the geological condition toward RC without using any prediction rating calculation. The higher of RC among both formulas are taken as the result of RC for selecting the support system.