Comparative analysis Machine learning(ML) based path loss dataset for university campus environment

Path loss is an important key factor affecting performance and efficiency in wireless communication system. Here primary measured data was collected from Covenant University, Ota, Ogun State, Nigeria & supplementary materials article found at the following address (https://doi.org/10.1016/j.dib.2018.02.026). In order to generalize for every university environment, four key features (Elevation, Altitude, Clutter height, Distance) were used to take as input, and a total of 3616 measured samples used in our cases to build the ML-based model and predict this dataset. Hence 28 different types of ML-based machine learning have been implemented to generate this predicted ML-based data. Each type is under categorized (ML) techniques such as Linear Regression, Tree, SVM (Support vector machine), Efficient Linear method, Ensemble method, GPR (Gaussian Process Regression), Neural Network and Kernel method.路径损耗（Path loss）是影响无线通信系统性能与运行效率的关键核心因素。本研究的原始实测数据采集自尼日利亚奥贡州奥塔市的圣约大学（Covenant University），补充材料可通过以下链接获取：https://doi.org/10.1016/j.dib.2018.02.026。为实现对各类大学校园环境的泛化适配，本研究选取高程（Elevation）、海拔（Altitude）、杂波高度（Clutter height）与传输距离（Distance）四项核心特征作为模型输入，并采用共计3616条实测样本构建基于机器学习（ML, Machine Learning）的模型，开展本数据集的预测任务。本研究共实现28种不同的基于机器学习的方法以生成该预测数据集，这些方法可归类为以下几类机器学习技术：线性回归（Linear Regression）、决策树（Tree）、支持向量机（SVM, Support Vector Machine）、高效线性方法、集成学习方法、高斯过程回归（GPR, Gaussian Process Regression）、神经网络以及核方法。