Font Size: a A A

Research On The Application Of Decision-Tree Method For Analysis Of College Examinees In Will-deciding Before Entrance Examination

Posted on:2008-08-28Degree:MasterType:Thesis
Country:ChinaCandidate:Z L LuoFull Text:PDF
GTID:2178360215472443Subject:Applied Mathematics
Abstract/Summary:PDF Full Text Request
Data mining is used to analyze data intensively, classify data clearly, gather data perfectly, find and depict data's trend accurately, mark abnormity definitely, find out useful information in data, and find undiscovered knowledge. By this, it could direct people's action and offer effective reference to business competition, production and management of corporations, government's decision-making and science exploration.Classification is one of the most familiar data mining's tasks. Using of the theory of information, the algorithm of decision tree's induction and classification could make the analysis of information content, find the important character of category-reflection, and seize the essence of problems exactly. It is an effective method of supervising learning. The notable point of the decision tree's induction and classification is that the effect is more ascendant for database with enormous samples. This is the academic base of its application on information management of the College Entrance Examination.By province as a unit, correlative notes of annual College Entrance Examination'wills information is more than millions, stored in multi-databases of every areas dispersedly. The data is enormous, containing abundant decision-making information and knowledge. Developing the precious information resource is one important task of serving the College Entrance Examination and instructing examinee of filling in wills scientifically. It is one of the problems which need settling urgently at present.Based on characters of the College Entrance Examination's wills information and taking pretreatment on it, the paper depicts of using C4.5 algorithm neatly for building binary decision tree based on unbalanced data-set, inosculating rules'extracting, and veracity evaluation and so on. It could make information-mining for the College Entrance Examination's wills information, get latent information of orderliness, build the forecast model of the College Entrance Examination's wills matriculate by linking of least squares method. By characters of grade, subject, requested specialty, school and so on about examinee, it could classify and forecast the College Entrance Examination'matriculate information, get correlative rules of the College Entrance Examination's matriculate which are provided to examinee for taking decision-analysis ulteriorly.Based on the above theory, the developed system called"the College Entrance Examination's wills information management system"has experimentized on data of the College Entrance Examination's wills information of Henan province of last three years. The result is inosculated with the distribution of the College Entrance Examination actual matriculate information of that year. Therefore, it is really sure that the testing results, as the effective reference value, are good to examinee of the coming year for filling in wills.
Keywords/Search Tags:data mining, analysis of college examinees in Will-deciding before entrance examination, unbalanced data-set, least squares method, C4.5 algorithm
PDF Full Text Request
Related items