Análisis de la fertilidad por medio de técnicas de minería datos

Q2 Multidisciplinary

Informacion Tecnologica Pub Date : 2022-06-01 DOI:10.4067/s0718-07642022000300203

O. Castrillón, J. A. Arango, Luis Fernando Castillo

引用次数: 1

Abstract

20%. Abstract The primary objective of this research study is to predict the most important variables that affect fertility in a person. The study is conducted by using the automatic learning and data mining platform Weka, the expectation maximization (EM) clustering algorithm, SimpleKMeans, and the classification algorithm J48, which behaves similarly to a Bayesian algorithm. Initially, an existing database is modeled until 105 records and nine variables are reached, eight independent variables (age, illnesses, accidents, surgeries, fever, alcohol, smoker, and sedentary lifestyle) and one dependent variable (fertility). The results revealed the five most influential variables: 1) age, 2) accidents, 3) fever, 4) surgery, and 5) alcohol. The success rate is over 90% when a cross-validation 80% - 20% is applied. It is concluded that the random forest and clustering algorithms employed here allow to clearly determine the most important variables that affect fertility in a person.

查看原文本刊更多论文

通过数据挖掘技术进行生育分析

20%。本研究的主要目的是预测影响一个人生育能力的最重要变量。本研究采用自动学习和数据挖掘平台Weka、期望最大化(EM)聚类算法、SimpleKMeans和行为类似贝叶斯算法的分类算法J48进行。最初，对现有数据库进行建模，直到得到105条记录和9个变量，8个自变量(年龄、疾病、事故、手术、发烧、酒精、吸烟和久坐的生活方式)和1个因变量(生育能力)。结果揭示了五个最重要的变量:1)年龄，2)事故，3)发烧，4)手术，5)酒精。当交叉验证80% - 20%时，成功率超过90%。结论是，这里使用的随机森林和聚类算法可以清楚地确定影响一个人生育能力的最重要变量。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Informacion Tecnologica Multidisciplinary-Multidisciplinary

自引率

0.00%

发文量

119

期刊介绍： The Información tecnológica magazine is a service of the Center for Information Technology (CIT), this service is restricted and prohibited their sale to third parties as well as the total or partial reproduction for commercial purposes. The articles presented in this magazine are for original papers sent by the authors and have been accepted for publication by a committee, and an Editorial Committee of Referees. The Center for Information Technology is not responsible for the opinions contained in the articles, that responsibility rests with the perpetrators of these.