共查询到20条相似文献，搜索用时 281 毫秒
1.
张洁玲 《福建工程学院学报》2017,(6):591-596
针对传统离群点检测算法的局限性进行研究,利用数据对象之间的相邻关系,提出了一种基于密度和距离相结合的离群检测算法,该算法解决了基于距离的离群检测算法不能准确识别局部离群点的问题,有效避免由于稀疏和密集簇过于邻近的而出现离群点误判的情况。通过在人工模拟数据及真实数据集上的实验测试证明改进算法的可行性,该算法能更有效地检测出数据集中的离群对象。 相似文献
2.
电类实验教学在高等院校中覆盖面广,实验过程中学生所测数据量大,人工评判数据工作繁琐。随着在线实验教学的普及,迫切需要一种实验数据智能评判的方法。依据Mean Shift思想提出一种基于距离的离群点检测(MSOD)算法,以理论计算值作为初始点,沿着概率密度梯度的方向寻找数据集最稠密的位置,与该位置的距离大于某一特定值的数据为离群点,离群点数据即为测量有误的数据。实验结果表明,MSOD算法识别错误实验数据的效果较好,可以有效地减少实验教学中重复繁琐的人工评判数据的工作,节约人力成本,提高实验教学效率。与现有的离群点检测算法比较,MSOD算法提高了错误数据识别的正确率,并且降低了时间成本。 相似文献
3.
K-means算法作为较为普遍的聚类算法,聚类效果受孤立点、噪声点和初始聚类中心影响较大。结合Isolation Forest算法计算数据中每个样本的异常度系数,根据离群值过滤比例计算得到异常度系数阈值,对高度异常值加以隔离,并对隔离后的数据集使用平均插值法求得初始聚类中心。运用改进K-means算法对真实数据集进行聚类分析,与此同时,通过比较多个离群值过滤比例下的聚类结果,找到离群值过滤比例的最优取值。仿真结果表明,相比于原始算法,新算法显著提升了聚类准确性,聚类效果更佳。 相似文献
4.
为了提高高维数据集合离群数据挖掘效率,在分析了传统的离群数据挖掘算法优点和缺点的基础上,提出了一种离群点检测算法,首先将非线性问题转化为高维特征空间中的线性问题,然后利用非线性数据变换进行维数约减,对从高维采样数据中恢复得到低维数据集,通过本文提出的离群数据假设,并结合本文给出的离群聚类方法对所得数据对象投影分量是否是离群数据进行判别。仿真实验的结果表明了该方法能够有效地发现高维数据集中的离群点。与此同时,该算法具有参数估计简单、参数影响不大等优点,为离群点检测问题的机器学习提供了一条新的途径。 相似文献
5.
学生评教工作的目的就是为了督促教师改进教学方法,提高教学质量.学生评价结果的客观公正是保证教学质量改进的关键一环.在全样本大数据背景下,利用数据挖掘的方法探索消除学生网络评教数据噪声方法,引导高校学生客观评价,保证评教数据的客观真实性,是高等教育教学一项亟待改进的工作.本文通过大数据背景下的数据挖掘对学生评教进行量化研究,找出学生评教规律,从而划分出学生评教噪声数据,研究降噪方法,减少噪声数据,促进学生评教工作的进一步规范,提高学生评教有效性. 相似文献
6.
《赤峰学院学报(自然科学版)》2016,(20)
离群点研究在实际应用中有着重要的意义,随着数据规模的不断扩大,传统的离群点检测方法已经不适用于高维空间数据,本文在遗传算法的基础上结合模拟退火算法,一方面利用遗传算法对高维数据处理有很好的全局搜索能力,一方面利用模拟退火算法的局部搜索能力,最后经实验证明,本文提出的新算法能有效的提高高维空间离群点检测的效率. 相似文献
7.
《实验室研究与探索》2020,(2):28-33
为降低风电场的运营成本和提高设备维护效率,提出了基于离群点检测和PSO-BP的风速预测模型。将基于距离和统计学的离群点检测方法结合,并通过分组剔除风速数据中的异常值;然后利用小波阈值去噪算法对风速数据进行去噪;最后使用粒子群算法优化后的BP神经网络进行预测。仿真结果证明,改进的离群点检测方法和小波阈值去噪降低了风速数据的波动性和随机性;对于3组不同风速数据,基于离群点检测和PSO-BP预测模型的预测精度均高于其他对比模型。 相似文献
8.
随着网络规模的不断扩大,为了保障网络安全、稳定运行,需要一种更高效的网络故障预警算法。通过对传统网络故障预警算法优缺点的分析,针对其缺点进行优化改进,采用离群点检测算法建立网络故障预警模型。对异常检测算法数据进行预处理,在 Hadoop 平台上计算数据异常指数,并不断调整阈值参数。实验结果表明,改进型离群点检测算法故障检测率达到 98%,可对网络故障进行有效预警。 相似文献
9.
一种改进的k-means聚类算法 总被引：2，自引：0，他引：2
针对k-means算法事先必须获知聚类数目以及难以确定初始中心的缺点,提出了一种改进的k-means聚类算法.首先引入轮廓系数的概念,通过计算不同K值下簇集中各对象的轮廓系数确定事先未知分类信息的数据集中所包含的最优聚类数Kopt;然后通过凝聚层次聚类的方法获得数据集的分布,确定初始聚类中心;最后利用传统的k-means方法完成聚类.理论分析表明,所提出的算法具有适度的计算复杂度.IRIS测试数据集的实验结果表明了该算法能够合理区分不同类型的簇集,且可以有效地识别离群点,聚合后的结果簇集具有较低的熵值. 相似文献
10.
以高校学生评教的原始记录为挖掘数据集,介绍了数据挖掘中关联规则算法在学生评教中的应用,学生评教结果与各类教师素质特征得到较客观的对应。 相似文献
11.
Student Evaluation of Teaching Effectiveness: An assessment of student perception and motivation 总被引：2，自引：2，他引：2
Over the past century, student ratings have steadily continued to take precedence in faculty evaluation systems in North America and Australia, are increasingly reported in Asia and Europe and are attracting considerable attention in the Far East. Since student ratings are the most, if not the only, influential measure of teaching effectiveness, active participation by and meaningful input from students can be critical in the success of such teaching evaluation systems. Nevertheless, very few studies have looked into students' perception of the teaching evaluation system and their motivation to participate. This study employs expectancy theory to evaluate some key factors that motivate students to participate in the teaching evaluation process. The results show that students generally consider an improvement in teaching to be the most attractive outcome of a teaching evaluation system. The second most attractive outcome was using teaching evaluations to improve course content and format. Using teaching evaluations for a professor's tenure, promotion and salary rise decisions and making the results of evaluations available for students' decisions on course and instructor selection were less important from the students' standpoint. Students' motivation to participate in teaching evaluations is also impacted significantly by their expectation that they will be able to provide meaningful feedback. Since quality student input is an essential antecedent of meaningful student evaluations of teaching effectiveness, the results of this study should be considered thoughtfully as the evaluation system is designed, implemented and operated. 相似文献
12.
13.
The Impact of Student Perceptions and Characteristics on Teaching Evaluations: A case study in finance education 总被引：2，自引：2，他引：2
Andrew C. Worthington 《Assessment & Evaluation in Higher Education》2002,27(1):49-64
This study uses an ordered probit model to examine the impact of student characteristics and perceptions of the teaching evaluation process on student ratings. The results indicate that expected grade, ethnic background, gender and age are a significant influence on student ratings. A primary student-based influence on teaching evaluation performance would appear to be the perceived potential outcome of the evaluation in terms of tenure, promotion and salary decisions, and improvements in teaching and staff allocation. The impact of student perceptions and characteristics is also found to vary across the various dimensions of teaching performance with the potential bias being highest for evaluation questions relating to overall performance, and lowest for questions relating to formative assessment and deep learning outcomes. 相似文献
14.
Ruth Neumann 《Assessment & Evaluation in Higher Education》2000,25(2):121-134
In the past decade student evaluation of teaching has expanded within Australian universities, with the results of evaluations increasingly being used to make judgements about teaching quality, career advancement and the funding of teaching. An important component of any student evaluation of teaching system is, therefore, the communication of rating results in a manner that enables fair and meaningful interpretations and comparisons of results by the wide range of evaluation users. This paper describes the development in one Australian university of Rating Interpretation Guides (RIGs) which take into account the influence of different teaching contexts on ratings and encourage evaluation users to explore ratings in terms of a range of scores rather than focusing on a single mean score. RIGs represent an important innovation in the communication of evaluation results for both formative and summative evaluation purposes. They are designed to enable meaningful interpretations of ratings which in turn will encourage appropriate use of results. 相似文献
15.
Teaching Effectiveness: The relationship between reaction and learning evaluation criteria 总被引：1，自引：0，他引：1
This study used a pre/post design to assess student learning for the purposes of examining relationships among student grades, student learning, and student evaluations of teaching. These relationships were reframed in terms of reaction (Level I) and learning (Level II) evaluation criteria. Participants were 652 undergraduate students enrolled in seven sections of an introductory psychology course. Our results indicated a medium relationship between student grades and the pre/post learning measure. In addition, a small relationship was observed between student ratings of teaching effectiveness and a pre/post measure of learning. We conclude that student ratings and learning measures assess different aspects of teaching effectiveness and should not be used interchangeably. The most appropriate criterion for assessing teaching effectiveness is a function of the goal of evaluation. However, reaction and learning measures may be used in conjunction to obtain a more complete picture of instructor effectiveness. 相似文献
16.
Craig S. Galbraith Gregory B. Merrill Doug M. Kline 《Research in higher education》2012,53(3):353-374
In this study we investigate the underlying relational structure between student evaluations of teaching effectiveness (SETEs)
and achievement of student learning outcomes in 116 business related courses. Utilizing traditional statistical techniques,
a neural network analysis and a Bayesian data reduction and classification algorithm, we find little or no support for the
validity of SETEs as a general indicator of teaching effectiveness or student learning. In fact, the underlying structure
appears to be non-linear and possibly negatively bimodal where the most effective instructors are within the middle percentiles
of student course ratings, while instructors receiving ratings in the top quintile or the bottom quintile are associated with
significantly lower levels of student achievement. 相似文献
17.
Student assessment of teaching in higher education 总被引：1，自引：0，他引：1
Allen H. Miller 《Higher Education》1988,17(1):3-15
Plans to introduce campus-wide assessments of college or university teaching which are largely dependent on student ratings are seen as a threat to academic freedom in those institutions with little or no experience of this form of evaluation. While regular student evaluations of teaching are very common in North America, their introduction is only now being considered in colleges and universities in a number of other countries. Research on the reliability and validity of student ratings indicate that they are capable of providing valuable information about the quality of teaching. Depending on the survey used, this type of evaluation may be used to provide evidence of teaching ability to staffing committees or to suggest ways of improving teaching. The paper concludes with a set of recommendations for higher education institutions which are considering the regular assessment of all teachers by their students. 相似文献
18.
Harry G. Murray 《Assessment & Evaluation in Higher Education》1984,9(2):117-132
This paper reviews recent research evidence on formative (developmental) and summative (judgemental) evaluation of teaching in North American universities. The major conclusions of the review are as follows: (1) student ratings are the most common method of evaluating teaching in North American universities, and their popularity appears to be increasing; (2) student ratings have been found to provide reliable and valid information on limited aspects of teaching competence; (3) teaching evaluations make a difference in decisions on faculty salary, tenure and promotion; (4) evaluation of teaching has been shown to produce modest improvements in teaching effectiveness; (5) faculty satisfaction with teaching evaluation practices ranges from very high to very low, depending on institutional context; and (6) there is consensus that several specific improvements are needed in the current teaching evaluation system. 相似文献
19.
Fadia Nasser-Abu Alhija Barbara Fresko 《Assessment & Evaluation in Higher Education》2018,43(6):943-954
Graduate teaching assistants (GTAs) constitute a valuable and economical teaching force in many higher education undergraduate programmes. However, student satisfaction with their teaching has attracted little attention in the research literature. This study aimed at examining students’ evaluation of teaching of GTAs in discussion groups, as well as exploring the effects of group and GTA variables on these ratings. Data were collected using a questionnaire administered online and completed by 7078 undergraduate students. Participants were enrolled in classes taught by 278 GTAs from four faculties in a major Israeli university. Results indicated that ratings assigned to clarity of instruction were the most salient predictor of students’ overall evaluation. Generally, findings were consistent with those reported in the literature for other categories of instructors. Groups taught by GTAs in exact sciences and engineering were rated higher than those in social sciences and business management. Group size and the percentage of men students were inversely correlated with student ratings, while student attendance rate was positively correlated. Women GTAs and GTAs who taught more than one group tended to receive higher ratings. Overall student attendance rate was the most prominent predictor of student ratings. The implications of the findings are discussed. 相似文献
20.
It is conventionally assumed that student ratings perform a significant function in driving improvement in pedagogical practices in higher education. As a result, this form of evaluation has gradually become institutionalised in recent decades as an essential proxy for understanding teaching and course quality in universities across the world. However, with the rise of market-based models in higher education and heightened expectations for accountability mechanisms, the role and functional purpose of ratings-based student evaluation have become increasingly confused. This rising ambiguity has created strong tensions between the seminal drive of student ratings as a tool of quality improvement, and the emerging demands for its use as a transparent accountability measure for the comparative assessment of academic performativity. So are student ratings now largely a tool of quality assurance or performance measurement, or do they remain a legitimate tool for pedagogical improvement? This paper reports on a study that responded to this critical question by considering the contemporary work of student ratings in a major Australian university. The outcomes of this research demonstrate that tension between improvement and accountability motives is causing considerable confusion and discord around the role and value of the student voice. It also reveals that academics are tending to discount the often critical insights of students on the implications of their pedagogical practices as a result of the elevating institutional role of student ratings as a proxy for teaching quality. In considering these outcomes, rising levels of academic dissonance around student ratings would suggest a necessity to consider broadened evaluative strategies that are able to more effectively capture the improvement potential offered by the student voice. 相似文献