Research Of Application Of Improved ABC-ELM And HMM In Student Evaluation System

Student evaluation,which is a kind of education evaluation that its evaluation object is student,is a process that educator measure students ' academic achievement,personality development,virtus state,physical quality and some other aspect by a certain standard and take the result as feedback on education pract:ice to improve the teaching process,and is an assessment of student's academic progresses and behavior changes.Due to the guiding role and educational role it plays,it is important for promoting students' growth and development.To be objective,as a efficient means of stimulating student to study,traditional student evaluation system and method have done their part of job as a guiding role,but there are still some defects in practice,such as,subjective opinion from expert can have a great impact on the distributing of weight of the indicator in traditional method,which leads to deficiency of objectivity,and some method of evaluation focus on the current state of student without referring to their past state,which can discharge them easily.Therefore,how to evaluate student in a more scientific and objective way becomes a hot topic.To solve the problem of assessment for student,this thesis study from the following two perspectives:weight distribution of evaluation indicators and learning ltrend evaluation.The study of weight distribution of evaluation indicators is about how to learn evaluation indicators weight distribution of sample data;which can supply support for setting evaluation indicators weight.The study of learning trend evaluation is about how to evaluate the learning trend of student based on the test scores of single subject in a period of time.To the problem of evaluation indicators weight distribution,the thesis proposes an evaluation indicators weight distribution method which based on ABC-ELM,this method use extreme learning machine to learn the distribution of indicators weight,meanwhile to the shortcome of ELM,we use improved ABC algorithm to optimize the choosing of ELM parameters.Compare to ELM,ABC-ELM improve the accuracy and stability of evaluation.To the problem of learning trend evaluation,the thesis proposes a learning trend evaluation method based on HMM,and a HMM training method using improved ABC algorithm.The thesis also designs some experiments to verify the effectiveness and practicality of both approach of evaluation.Eventually the thesis designs and implements the middle school student evaluation system based on the method of weight distribution of evaluation indicators and learning trend evaluation.