Due to space limitations in urban areas, underground construction has become a common practice worldwide. When using deep excavations, excessive lateral movements of wall and adjacent structure are a major concern because they can lead to significant displacements in adjacent structures. Therefore, an accurate predictions of lateral wall deflections were a major concern for design and construction of excavation support system. The objective of this thesis is to develop a simple polynomial regression model for predicting maximum wall deflection due to deep excavation, in clays. The excavation width, excavation depth, soil shear strength, wall embedment depth and wall thickness were varied to generate maximum wall deflection and to study the wall deflection behavior. Eight excavation case histories of deep excavation in thick layers of clays was adopted for generation of data used for development of intended model as a basis of numerical analyses.rnThis research used extensive plane strain (2D) analysis to examine the wall deflection for excavation-induced ground deformation in clay deposits supported by diaphragm wall and bracing, using the Finite Element Method (FEM).rnFor the development of the Simple Polynomial Regression models for estimating the maximum wall deflection, a database of excavation case histories are compiled and to complement the database of excavation case histories, a large number of data was generated using FEM solutions, by varying the principal parameters considered in the study including excavation depth, excavation width, soil shear strength, embedment depth and wall stiffness with Mohr coulomb soil model. Based on the established database, a Simple Polynomial Regression model is developed for predicting the maximum wall deflection, caused by braced excavations in clays. The proposed method provides a fairly good prediction of wall deflection as compared to FEM.rnA Simple Polynomial Regression models has been proposed to estimate the lateral deflection of wall braced excavation in clay with average to good workmanship. Computed values by SPRM (simple polynomial regression model) agreed well with results from observed maximum wall deflection from different excavation case histories which are not used in the development of simple model.