Accuracy and precision 意义

前言

在做数据分析时常常用到两个概念:精度(Precision)和准确度(Accuracy)。这两个词可以在口语中作为同义词使用,但它们在科学方法背景下是完全不同的。其具体意义是怎样的,在科学研究中需要深究,以免报告交流时产生错用的尴尬。

来自Wiki中的定义

Accuracy:指在一定实验条件下多次测定的平均值与真值相符合的程度,以误差来表示,用来表示系统误差的大小。

Precision:是随机误差的描述,是统计变异性的度量。是多次重复测定同一量时各测定值之间彼此相符合的程度。表征测定过程中随机误差的大小。

ISO 5725-1,定义“Accuracy”用于描述测量与真实值的接近程度。 当该术语被应用于相同测量的测量集时,它涉及随机误差的分量和系统误差的分量。 在这种情况下,actual (true)值是一组测量结果的均值与实际(真实)值的接近程度,精度是一组结果之间的一致性。

简单说,给定一组测量的数据点(比如多次测量一段直线的长度),如果这些值接近被测量的平均值,则可以说这组数据是精确的;如果这组数据接近被测量的真实值,则可以说测量是精确的。这两个概念是彼此独立的,因此一组特定的数据可以说是准确的,或是精确的,或两者都是,或两者都不是。


这里写图片描述
图:Accuracy是测量结果与真实值的接近度,precision是重复性测量误差。

进一步解释

在科学,工程和统计领域,测量系统的准确度是数次测量值与该真实值的接近程度。测量系统的精度与重现性和重复性有关,是在不变条件下重复测量结果相同的程度。

测量系统可以是准确的,但不精确,也可以是精确但不准确的,··· ···。例如,如果实验包含系统误差,则增加样本数量通常会提高精度,但不会提高准确度。消除系统误差可以提高准确度,但不会改变精度。如果一个测量系统既准确又精确,则被认为是有效的。

除了精度和精度外,测量还可能具有测量分辨率。测量分辨率是在测量中响应物理量的最小变化的度量,可以用精度表示。


这里写图片描述
这四个射击点点阐述了accurate和precise的区别

参考

https://en.wikipedia.org/wiki/Accuracy_and_precision
http://blog.csdn.net/luo123n/article/details/50754104
http://f.dataguru.cn/forum.php?mod=viewthread&action=printable&tid=708550
https://celebrating200years.noaa.gov/magazine/tct/tct_side1.html

"Automatic Learning Based on Multi-Modal Data for High-Accuracy Learning Style Recognition" I. Introduction A. Research Background The study of learning styles has been a popular topic in the field of education for many years. It is believed that understanding an individual's learning style can lead to more effective teaching methods and better educational outcomes. B. Research Objectives This research aims to investigate the feasibility of using multi-modal data to accurately recognize a person's learning style. The study seeks to identify the most effective combination of modalities for recognizing learning styles and to compare the results with previous research. C. Research Significance The results of this study could contribute to the development of more personalized learning environments and provide valuable information for educators and trainers. II. Literature Review A. Overview of Learning Styles Learning styles refer to the way individuals process and understand information. There are various models of learning styles, but most can be categorized into visual, auditory, and kinesthetic. B. Previous Research on Learning Style Recognition Previous research has focused on recognizing learning styles through self-report surveys or by observing the individual's behavior in a learning environment. However, these methods have limitations and may not provide accurate results. C. Overview of Multi-Modal Data Multi-modal data refers to data that is collected through multiple sources or modalities. This type of data is becoming increasingly prevalent in the digital age and can provide a more comprehensive representation of a person. D. Multi-Modal Data Integration and Learning Style Recognition The integration of multi-modal data can provide a more complete picture of an individual's learning style, leading to improved recognition accuracy. This research will explore the use of multi-modal data for learning style recognition. III. Methodology A. Data Collection The data for this study will be collected from multiple sources, including self-report surveys, physiological measurements, and behavioral observations. B. Data Preprocessing The collected data will undergo preprocessing to ensure it is suitable for analysis. This may include cleaning, normalizing, and transforming the data. C. Feature Extraction Features will be extracted from the preprocessed data to represent the most important characteristics for learning style recognition. D. Model Selection The most appropriate model for recognizing learning styles will be selected based on the extracted features and evaluated using appropriate performance metrics. E. Model Evaluation The performance of the selected model will be evaluated using appropriate metrics, such as accuracy, precision, and recall. The results will be compared with previous research to determine the effectiveness of the multi-modal approach. IV. Results A. Results Overview The results of this study will provide insight into the feasibility of using multi-modal data for learning style recognition. The results will also provide information on the most effective combination of modalities and the impact of each modality on recognition accuracy. B. Comparison with Previous Research The results will be compared with previous research to determine the effectiveness of the multi-modal approach and to identify areas for improvement. C. Analysis of Modality Impact The impact of each modality on recognition accuracy will be analyzed to determine the most valuable sources of information for learning style recognition. D. Discussion of Results The results will be discussed in the context of their implications for education and their limitations. The discussion will also provide recommendations for future research in this area. V. Conclusion A. Summary of Findings The results of this study will provide valuable information on the feasibility of using multi-modal data for learning style recognition. The findings will also provide insights into
评论 2
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值