A Study of The Survey of Family Income and Expenditure by Using Data Mining Technology
| Autor: | HUNG, SHIH-TING, 洪詩婷 |
|---|---|
| Rok vydání: | 2017 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 105 The different family characteristics may cause uneven distribution of resources. And then affect the family consumption ability and expenditure patterns. That represents family consumption ability is the reflective of living standard. With the age of big data, data mining has become an indispensable technology. How to find out useful information from huge amount of data and effectively reduce the computing time that is the key points during search we want to know. In this study, we selected information from the Survey of Family Income and Expenditure from 2009 to 2015 as research data. To discuss the Survey of Family Income and Expenditure in depth. We use three kinds of data mining technology to build models. One is decision tree, another is neural network and the other is random forest. To find the different family consumption and head of the household economic characteristic belong to what kinds of the consumption expenditure pattern. We also predict the accuracy of these three kinds of models. In result, we discover family structures and family disposable income are important in independent variable. Income earners population, gender of head of the household and house belongings are less important in independent variable. Residential services, utilities and other fuels consumption, health care consumption, alcohol and tobacco and betel nut consumption and miscellaneous consumption are less important in dependent variable. The model comparison results show that neural network is the most accurate prediction model. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|