In this work, we propose a machine learning-based approach to predict the efficiency of the semiconductor neutron detector for planar configuration detector design at different Low-Level Discrimination (LLD) value setting for boron carbide converter material. Prior to the fabrication of a semiconductor neutron detector, it is pertinent to optimize several parameters relevant to detector design. Monte Carlo techniques are used for this purpose. GEANT4 toolkit, which utilizes the Monte Carlo approach, is commonly utilized to estimate these parameters. A noteworthy limitation of the GEANT4 simulation is its high computational complexity. The simulation demands substantial computational resources and can be computationally demanding. As a consequence, the processing times can become lengthy, and this may pose challenges in conducting simulations on a larger scale or with intricate systems. Therefore, in this study, we adopted a different novel approach by leveraging machine learning techniques to predict the efficiency of a planar semiconductor neutron detector under varying settings of the LLD values. The efficiency values obtained from the GEANT4 simulations formed the basis of the dataset used in this study. Subsequently, different machine learning techniques, such as Linear Regression, Polynomial Regression, Support Vector Regression (SVR), and Neural Networks were explored in order to develop an optimized model. The aim was to accurately predict efficiency values by considering the influence of converter layer thickness and different LLD settings. Initially, we trained these models by utilizing the GEANT4 dataset, specifically focusing on efficiency up to 700 keV LLD. Subsequently, we employed the trained model to predict the efficiency for LLD values of 800 keV and 900 keV. Through our analysis, we determined that the neural network emerged as the most accurate model for predicting efficiency, closely aligning with the efficiency simulated by GEANT4 software toolkit.