Research On Optimization Algorithm Of Cognitive Learning Quality Evaluation Based On Cloud Computing

Topics of this article come from the humanities and Social Sciences Department of Liaoning province "mobile cloud computing environment based on cognitive modeling and optimization model of teaching evaluation in colleges and universities" project.Research is the starting point for an effective auxiliary teaching in a cloud computing environment evaluation models.This paper focuses on in-depth study of quality evaluation of learning.Evaluation of system design evaluation system based on workflow and platform is divided into four modules are: data acquisition,data storage,data analysis and data query module.Collection of data through a data acquisition module,after data storage module for data storage,data preparation module for data analysis,data analysis,users can use the data module queries the user's result.Is one of the most important data analysis module,that module by way of improved algorithms make the results better.This study is divided into two main areas: questionnaire items were selected on the one hand,on the other hand is prediction algorithms.In the questionnaire of learning quality evaluation based on the cognitive and the establishment,from the cognitive learning theory,using internationally recognized the motivated strategies for learning questionnaire(MSLQ)on the basis of and in connection with the evaluation of study quality: the SOLO taxonomy theory(learning structures that can be observed)changes from the book.Will be received in the first round of survey data using SPSS for validity and reliability testing out inapplicable items,keep on learning and influencing the quality of items.Testing of internal consistency from previous 0.926 to 0.930.Description the high reliability of the questionnaire,as a basis for quality evaluation based on cognitive learning.For prediction of return data is obtained through questionnaire return recycling.On cloud computing platforms,using the Python language to write the corresponding algorithms and regression for numeric data.Regression will be facing two problems,a poor fitting,and the second is expansion bottleneck.This research to more than one independent variable dependent variable,so first by multiple linear regression method.Purpose of regression is to obtain better results,due to the multiple linear regressionto determine the regression model is to meet all of the sample rules tend to also take into account the outlier model,the model will appear in the end poor fitting.In order to improve prediction performance,improvement of basic algorithms for local weights,locally weighted linear regression is used to improve forecast accuracy.Along with the in-depth study,the amount of data will continue to surge,extensions will appear in the data analysis bottleneck problems,in order to enable an improved algorithm on the operation rate has been raised,locally weighted linear regression algorithm for parallel processing,make them consistent with operational requirements of the MapReduce programming model.Finally,a data analysis of the multiple correlation coefficient r,and the size of the sum of squares of the residuals to determine the precision of the model.Users through the prediction of quality evaluation system for users after learning to do planning in a timely manner.