Decision tree support vector machine based on genetic algorithm for multi-class classification

To solve the multi-class fault diagnosis tasks, decision tree support vector machine （DTSVM）, which combines SVM and decision tree using the concept of dichotomy, is proposed. Since the classification performance of DTSVM highly depends on its structure, to cluster the multi-classes with maximum distance between the clustering centers of the two sub-classes, genetic algorithm is introduced into the formation of decision tree, so that the most separable classes would be separated at each node of decisions tree. Numerical simulations conducted on three datasets compared with ＂one-against-all＂ and ＂one-against-one＂ demonstrate the proposed method has better performance and higher generalization ability than the two conventional methods.

Use of Discrete Wavelet Features and Support Vector Machine for Fault Diagnosis of Face Milling Tool

This paper presents the fault diagnosis of face milling tool based on machine learning approach.While machining,spindle vibration signals in feed direction under healthy and faulty conditions of the milling tool are acquired.A set of discrete wavelet features is extracted from the vibration signals using discrete wavelet transform(DWT)technique.The decision tree technique is used to select significant features out of all extracted wavelet features.C-support vector classification(C-SVC)andν-support vector classification(ν-SVC)models with different kernel functions of support vector machine(SVM)are used to study and classify the tool condition based on selected features.From the results obtained,C-SVC is the best model thanν-SVC and it can be able to give 94.5%classification accuracy for face milling of special steel alloy 42CrMo4.

Credit Card Fraud Detection Using Machine Learning Techniques

Credit card companies must be able to identify fraudulent credit card transactions so that clients are not charged for items they did not purchase. Previously, many machine learning approaches and classifiers were used to detect fraudulent transactions. However, because fraud patterns are always changing, it is becoming increasingly vital to investigate new frauds and develop the model based on the new patterns. The purpose of this research is to create a machine learning classifier that not only detects fraud but also detects legitimate transactions. As a result, the model should have excellent accuracy, precision, recall, and f1-score. As a result, we began with a large dataset in this study and used four machine learning classifiers: Support Vector Machine (SVM), Decision Tree, Naïve Bayes, and Random Forest. The random forest classifier scored 99.96% overall accuracy with the best precision, recall, f1-score, and Matthews correlation coefficient in the experiments.

基于机器学习的冠心病风险预测模型构建与比较

背景冠状动脉粥样硬化性心脏病(以下简称冠心病)是全球重要的死亡原因之一。目前关于冠心病风险评估的研究在逐年增长。然而,在这些研究中常忽略了数据不平衡的问题,而解决该问题对于提高分类算法中识别冠心病风险的准确性至关重要。目的探索冠心病的影响因素,通过使用2种平衡数据的方法,基于5种算法建立冠心病风险相关的预测模型,比较这5种模型对冠心病风险的预测价值。方法基于2021年美国国家行为风险因素监测系统(BRFSS)横断面调查数据筛选出112606名研究对象的健康相关风险行为、慢性健康状况等24个变量信息,结局指标为自我报告是否患有冠心病并据此分为冠心病组和非冠心病组。通过进行单因素分析和逐步Logistic回归分析探索冠心病发生的影响因素并筛选出纳入预测模型的变量。随机抽取112606名受访者的10%(共计11261名),以8∶2的比例随机划分为训练与测试的数据集,采用随机过采样和合成少数过采样技术(SMOTE)两种过采样的方法处理不平衡数据,基于k最邻近算法(KNN)、Logistic回归、支持向量机(SVM)、决策树和XGBoost算法分别建立冠心病预测模型。结果两组年龄、性别、BMI、种族、婚姻状态、教育水平、收入水平、家里有几个孩子、是否被告知患高血压、是否被告知处于高血压前期、是否被告知患妊娠高血压、现在是否在服用高血压药物、是否被告知患有高脂血症、是否被告知患有糖尿病、吸烟情况、过去30 d内是否至少喝过1次酒、是否为重度饮酒者、是否为酗酒者、过去30 d内是否有体育锻炼、心理健康状况以及自我健康评价比较,差异有统计学意义(P<0.05)。逐步Logistic回归分析结果显示:年龄、性别、BMI、种族、教育水平、收入水平、是否被告知患高血压、是否被告知处于高血压前期、是否被告知患妊娠高血压、现在是否在服用高血压药物、是否被告知患有高脂血症、是否被告知患有糖尿病、吸烟情况、过去30 d内是否至少喝过1次酒、是否为重度饮酒者、是否为酗酒者以及自我健康评价为冠心病的影响因素(P<0.05)。风险模型构建的分析结果显示:k最邻近算法、Logistic回归、支持向量机、决策树和XGBoost采用SMOTE处理不平衡数据的总体分类精度分别为59.2%、67.4%、66.2%、69.2%和85.9%,召回率分别为75.2%、71.4%、70.5%、62.9%和34.8%,精确度分别为15.4%、18.2%、17.5%、17.6%和28.7%,F值分别为0.256、0.290、0.280、0.275和0.315,受试者工作特征曲线下面积分别为0.80、0.78、0.72、0.72和0.82;采用随机过采样处理不平衡数据的总体分类精度分别为62.5%、68.5%、69.0%、60.2%和70.1%,召回率分别为70.0%、69.5%、71.9%、69.0%和67.6%;精确度分别为15.8%、18.4%、19.1%、14.8%和19.0%,F值分别为0.258、0.291、0.302、0.244和0.297,受试者工作特征曲线下面积分别为0.80、0.77、0.72、0.72和0.83。结论本研究不仅确认了已知冠心病的影响因素,还发现了自我健康评价水平、收入水平和教育水平对冠心病具有潜在影响。在使用2种数据平衡方法后,5种算法的性能显著提高。其中XGBoost模型表现最佳,可作为未来优化冠心病预测模型的参考。此外,鉴于XGBoost模型的优异性能以及逐步Logistic回归的操作便捷和可解释性,推荐在冠心病风险预测模型中结合使用数据平衡后的XGBoost和逐步Logistic回归分析。

Application of four machine-learning methods to predict short-horizon wind energy

Renewable energy has garnered attention due to the need for sustainable energy sources.Wind power has emerged as an alternative that has contributed to the transition towards cleaner energy.As the importance of wind energy grows,it can be crucial to provide forecasts that optimize its performance potential.Artificial intelligence(AI)methods have risen in prominence due to how well they can handle complicated systems while enhancing the accuracy of prediction.This study explored the area of AI to predict wind-energy production at a wind farm in Yalova,Turkey,using four different AI approaches:support vector machines(SVMs),decision trees,adaptive neuro-fuzzy inference systems(ANFIS)and artificial neural networks(ANNs).Wind speed and direction were considered as essential input parameters,with wind energy as the target parameter,and models are thoroughly evaluated using metrics such as the mean absolute percentage error(MAPE),coefficient of determination(R~2),and mean absolute error(MAE).The findings accentuate the superior performance of the SVM,which delivered the lowest MAPE(2.42%),the highest R~2(0.95),and the lowest MAE(71.21%)compared with actual values,while ANFIS was less effective in this context.The main aim of this comparative analysis was to rank the models to move to the next step in improving the least efficient methods by combining them with optimization algorithms,such as metaheuristic algorithms.

Machine Learning and Artificial Neural Network for Predicting Heart Failure Risk

Heart failure is now widely spread throughout the world.Heart disease affects approximately 48%of the population.It is too expensive and also difficult to cure the disease.This research paper represents machine learning models to predict heart failure.The fundamental concept is to compare the correctness of various Machine Learning(ML)algorithms and boost algorithms to improve models’accuracy for prediction.Some supervised algorithms like K-Nearest Neighbor(KNN),Support Vector Machine(SVM),Decision Trees(DT),Random Forest(RF),Logistic Regression(LR)are considered to achieve the best results.Some boosting algorithms like Extreme Gradient Boosting(XGBoost)and Cat-Boost are also used to improve the prediction using Artificial Neural Networks(ANN).This research also focuses on data visualization to identify patterns,trends,and outliers in a massive data set.Python and Scikit-learns are used for ML.Tensor Flow and Keras,along with Python,are used for ANN model train-ing.The DT and RF algorithms achieved the highest accuracy of 95%among the classifiers.Meanwhile,KNN obtained a second height accuracy of 93.33%.XGBoost had a gratified accuracy of 91.67%,SVM,CATBoost,and ANN had an accuracy of 90%,and LR had 88.33%accuracy.

Ground Ozone Level Prediction Using Machine Learning

Because of the increasing attention on environmental issues, especially air pollution, predicting whether a day is polluted or not is necessary to people’s health. In order to solve this problem, this research is classifying ground ozone level based on big data and machine learning models, where polluted ozone day has class 1 and non-ozone day has class 0. The dataset used in this research was derived from the UCI Website, containing various environmental factors in Houston, Galveston and Brazoria area that could possibly affect the occurrence of ozone pollution [1]. This dataset is first filled up for further process, next standardized to ensure every feature has the same weight, and then split into training set and testing set. After this, five different machine learning models are used in the prediction of ground ozone level and their final accuracy scores are compared. In conclusion, among Logistic Regression, Decision Tree, Random Forest, AdaBoost, and Support Vector Machine (SVM), the last one has the highest test score of 0.949. This research utilizes relatively simple methods of forecasting and calculates the first accuracy scores in predicting ground ozone level;it can thus be a reference for environmentalists. Moreover, the direct comparison among five different models provides machine learning field an insight to determine the most accurate model. In the future, Neural Network can also be utilized to predict air pollution, and its test scores can be compared with the previous five methods to conclude the accuracy of Neuron Network.

Applied Machine Learning Methods for Detecting Fractured Zones by Using Petrophysical Logs

In the last decade, a few valuable types of research have been conducted to discriminate fractured zones from non-fractured ones. In this paper, petrophysical and image logs of eight wells were utilized to detect fractured zones. Decision tree, random forest, support vector machine, and deep learning were four classifiers applied over petrophysical logs and image logs for both training and testing. The output of classifiers was fused by ordered weighted averaging data fusion to achieve more reliable, accurate, and general results. Accuracy of close to 99% has been achieved. This study reports a significant improvement compared to the existing work that has an accuracy of close to 80%.

The Design of Predictive Model for the Academic Performance of Students at University Based on Machine Learning

Students in South African Universities come from different socio-cultural backgrounds, countries and high schools. This suggests that these students have different experiences which impact on their levels of grasping information in class as they potentially use different lenses on tuition. The current practice in Universities in contributing to the academic performance of students includes the use of tutors, the use of mobile devices for first year students, use of student assistants and the use of different feedback measures. What is problematic about the current practice is that students are quitting university in high numbers. In this study, knowledge has been drawn from data through the use of machine learning algorithms. Bayesian networks, support vector machines （SVMs） and decision trees algorithms were used individually in this work to construct predictive models for the academic performance of students. The best model was constructed using SVM and it gave a prediction of 72.87% and a prediction cost of 139. The model does predict the performance of students in advance of the year-end examinations outcome. The results suggest that South African Universities must recognize the diversity in student population and thus provide students with better support and equip them with the necessary knowledge that will enable them to tap into their full potential and thus enhance their skills.

基于Stacking算法集成学习的页岩油储层总有机碳含量评价方法

总有机碳含量(TOC)是页岩油储层评价的重要参数,而传统总有机碳含量测井评价方法精度较低且普适性较差,机器学习模型在一定程度上提高了总有机碳含量预测精度,但结果存在不稳定性。为了进一步提高页岩油储层总有机碳含量预测精度,基于有机质岩石物理特征和不同总有机碳含量测井响应特征的深入分析,优选出深侧向电阻率、声波时差、补偿中子和密度测井曲线作为总有机碳含量的敏感测井响应,并将其作为输入特征,以岩心分析总有机碳含量作为期望输出值,分别建立了决策树模型、支持向量回归机模型、BP(Back Propagation)神经网络模型,并建立了以决策树模型为基模型、支持向量回归机模型为元模型的Stacking算法集成学习模型。利用B油田A区块的岩心样本数据和实际井数据对不同模型预测总有机碳含量结果进行了验证,结果表明,基于Stacking算法的集成学习模型的总有机碳含量预测精度最高,相较于决策树模型、支持向量回归机模型、BP神经网络模型和改进的ΔlgR法,预测精度有较大提高。因此,基于Stacking算法的集成学习模型为该研究区最有效的总有机碳含量计算方法,这为准确地评估页岩油储层的生烃潜力、确保页岩油储层的高效开采及资源利用奠定了基础。

基于3种机器学习算法构建宫颈癌术后尿潴留风险预测模型

目的:运用决策树、逻辑回归和支持向量机构建宫颈癌根治性切除术后尿潴留风险预测模型并比较性能,为评估及预防宫颈癌术后尿潴留提供参考依据。方法:回顾性收集459例宫颈癌根治性切除术病人的临床资料,采用决策树、支持向量机和逻辑回归3种机器学习方法构建宫颈癌根治性切除术后尿潴留风险预测模型,采用准确性、召回率、精确率、F1指数和受试者工作特征(ROC)曲线下面积(AUC)评价模型性能。结果:共纳入病人的年龄、疾病分期、体质指数等8个变量。选择80%的数据集(367例)作为训练集,20%的数据集(92例)作为验证集,结果显示,决策树在训练集和验证集中准确率、召回率、精确率、F1指数和AUC都比支持向量机和逻辑回归更优,说明决策树在构建宫颈癌术后尿潴留风险预测模型中具有较高的准确率及较好的泛化性能;支持向量机在训练集中准确率、召回率、精确率、F1指数和AUC都比逻辑回归更优。同时,在验证集中,支持向量机的召回率和F1指数比逻辑回归更优,但是支持向量机的准确率、精确率和AUC却比逻辑回归差,说明支持向量机在宫颈癌术后尿潴留数据集中的泛化能力比逻辑回归差。结论:决策树在构建宫颈癌根治性切除术后尿潴留风险预测模型中具有较高的性能及较好的泛化能力,可为相关临床决策提供指导建议。

基于时频域分析方法与分类器算法相结合的Shunt辨识

为了准确辨识汽车抖动和半轴扭矩振荡(Shunt)现象,以传统汽车为载体,结合短时傅里叶变换方法分析补充了Shunt的定义并优化标签数据集,使用决策树、支持向量机和随机森林算法,以发动机转速、变速器输入轴转速等传感器信号作为输入来识别Shunt。结果表明,与传统的模型构建方法相比,该方法降低了模型构建的难度和成本,后续可用于可解释的机器学习来解释模型。

基于Google Earth Engine的前郭县春季农田覆膜提取

本文基于Google Earth Engine(GEE)云平台,综合考虑光学影像的波段反射率、光谱指数特征和雷达影像的极化、纹理特征,分别构建仅使用光学特征、仅使用雷达特征以及光学和雷达特征组合3种特征输入组合;根据精度确定最佳输入特征后,分别结合机器学习中的分类与回归树、支持向量机、最小距离分类法、梯度提升树和随机森林5种方法建立覆膜提取模型,依据结果精度评估不同方法的性能,并基于最优化模型提取出最终的覆膜农田面积。结果表明:①最佳输入特征为波段反射率特征+光谱指数特征+极化特征+纹理特征;②采用随机森林方法建立的模型精度最高,研究区I的总体精度达到了95.84%,Kappa系数为0.95,地物错分率为1.2%,明显优于其他4种方法(地物错分率较分类与回归树、支持向量机、最小距离和梯度提升树法降低0.8%、7.3%、38.0%和0.3%),研究区II的验证精度达到了87.84%,证明该模型在覆膜提取中可以取得更加准确的结果;③使用本文方法得到2022年研究区I覆膜农田面积为1302.48 km2,估算地膜使用量约为7585.62 t。本文综合考虑光学和雷达影像在地物识别中的特点建立模型,可以准确、高效的识别农田地膜,掌握地膜面积,对环境治理与防治具有重要意义。

肺恶性肿瘤内科诊断组DRG分组方案研究

目的:分析肺恶性肿瘤内科诊断组住院费用的影响因素,设计DRG分组方案,为研究对象分组方案优化提供案例分析和参考。方法:收集洛阳市某三甲医院2019—2022年肺恶性肿瘤内科诊断组的患者住院信息,采用K-means聚类和支持向量机分析住院费用的影响因素,通过CHAID算法构建DRG分组方案。结果:治疗方式、住院天数被纳入分组模型,最终生成6个DRG组,各DRG组的组内一致性好,组间差异性显著,分组效果好。结论:对于肺恶性肿瘤内科诊断组,住院天数分组效果好,但不适合作为分组节点;治疗方式有助于完善研究对象的DRG分组,但其划分方案有待进一步研究。

基于机器学习的传感网核心节点漏洞检测仿真

传感网的核心节点具有能量受限、难补给的特点,导致节点轮休时易出现的覆盖漏洞问题,造成传感网监测盲区。为此提出基于机器学习的传感网核心节点漏洞检测方法。利用支持向量机树形多分类器获取核心节点的位置。采取主成分分析法提取核心节点特征,将其输入到LSTM长短记忆神经网络模型中,并利用滑动窗口与哈希函数训练漏洞检测分类模型,完成传感网核心节点的漏洞检测。实验结果表明,研究方法检测传感网漏洞时平均耗时为13.6ms,检测率和准确率均可高达95%,计算得到性能消耗低于10%,90%的用户响应时间均在50ms以内。

一种基于遗传算法的SVM决策树多分类策略研究

提出了一种基于遗传算法(GA)的SVM最优决策树生成算法,并将其应用于解决SVM多分类问题.首先以最大分类间隔为准则,利用遗传算法对传统的SVM决策树进行优化,生成最优(或近优)决策二叉树;然后在各个决策节点,利用传统的SVM算法进行二值分类,最终实现SVM的多值分类.理论分析及实验结果表明,新方法比传统的DT-SVM、DAG-SVM方法有更高的分类精度,比经典的1-a-1、1-a-r有更高的训练和分类效率.

基于Stacking元学习策略的电力系统暂态稳定评估

为提高电力系统暂态稳定评估单个模型的准确率,研究了基于元学习策略的暂态稳定评估问题,提出了支持向量机、决策树、朴素贝叶斯和K最近邻法作为基学习算法,线性回归为元学习算法的Stacking评估模型。该模型将上述基学习算法的概率输出作为新训练数据的输入特征,同时保留原始的类标识。线性回归算法在新训练集上学习得到最终暂态稳定评估结果。新英格兰39节点测试系统和IEEE50机测试系统上仿真实现了该模型,仿真结果证明所提模型比单个模型的评估性能更好,为电力系统暂态稳定评估提供了新的思路。

基于粒子群算法的决策树SVM多分类方法研究

针对SVM多分类问题提出了一种基于粒子群算法的最优决策树SVM生成算法,以解决传统支持向量机多分类方法存在的不可分区域和误差积累现象。该方法利用自变异的PSO聚类算法在每一决策节点自动寻找最优或近优分类决策,将数据集划分为两类,直至叶子节点为止,最终根据最优决策树构建SVM多分类结构,训练各个节点SVM分类器。将该算法应用于图像人群密度分类问题,仿真实验表明,分类精度和分类时间得到明显改善,是一种有效地的多分类算法。

基于决策支持向量机的中文网页分类器

提出了基于决策支持向量机的中文网页分类算法。把支持向量机方法和二叉决策树的基本思想结合起来构成多类别的分类器，用于中文网页分类，从而减少支持向量机分类器训练样本的数量，提高训练效率。实验表明，该方法训练数据规模大大减少，训练效率较高，同时具有较好的精确率和召回率。

基于室外图像的天气现象识别方法

为提高室外视频监控的准确率,实现天气现象的自动观测,提出了一种基于室外图像的天气现象识别方法,该方法通过分析天气现象对图像的影响,提取图像功率谱斜率、对比度、噪声和饱和度等特征进行训练与分类,在训练过程中根据类别之间的特征距离建立分类决策树,并为决策树上非叶子节点构造支持向量机(SVM)分类器,并在每个分类器构造过程中通过对特征赋权值实现对特征的选择。通过对WILD图像数据库和采集图像集不同天气800个样本的测试,除了对降雨的识别率较低(75%)外,对晴、阴、雾天气的识别率均高于85%。