Evaluation woes: Metrics can help beat bias
نویسندگان
چکیده
منابع مشابه
How can software metrics help novice programmers?
Many computing education studies have reported poor learning outcomes in programming courses for novices. Yet methods for measuring students’ ability to generate computer programs remains an open research problem. In this paper we review some limitations of existing approaches for assessing student programs. We then propose a set of valid and reliable metrics for the direct measurement of novic...
متن کاملCan online trading algorithms beat the market? An experimental evaluation
From experimental evaluation, we reasonably infer that online trading algorithms can beat the market. We consider the scenario of trading in financial market and present an extensive experimental study to answer the question “Can online trading algorithms beat the market?”. We evaluate the selected set of online trading algorithms on DAX30 and measure the performance against buy-and-hold strate...
متن کاملCan we beat influenza?
or the past 65 years, the Global Influenza Surveillance and Response System (GISRS), coordinated by the World Health Organization (WHO), has engaged in open and efficient sharing of information, viruses, and responsibilities. The GISRS’s extraordinary longevity can be attributed to several generations of dedicated scientists and to the engagement of over 100 countries, often with limited resour...
متن کاملEvaluation Metrics in Classiication: a Quantiication of Distance-bias
This paper provides a characterization of bias for evaluation metrics in classiica-tion (e.g., Information Gain, Gini, 2 , etc.). Our characterization provides a uniform representation for all traditional evaluation metrics. Such representation leads naturally to a measure for the distance between the bias of two evaluation metrics. We give a practical value to our measure by observing if the d...
متن کاملA Quantiication of Distance-bias between Evaluation Metrics in Classiication
This paper provides a characterization of bias for evaluation metrics in classiication (e.g., Information Gain, Gini, 2 , etc.). Our characterization provides a uniform representation for all traditional evaluation metrics. Such representation leads naturally to a measure for the distance between the bias of two evaluation metrics. We give a practical value to our measure by observing if the di...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Nature
سال: 2018
ISSN: 0028-0836,1476-4687
DOI: 10.1038/d41586-018-05751-4