عنوان المقالة:MMS: Minimum Maximum Strategy for Classification and Testing MMS: Minimum Maximum Strategy for Classification and Testing
ا.د. محمد عصام يونس | Mohammed I. Younis | 14092
- نوع النشر
- مجلة علمية
- المؤلفون بالعربي
- محمد عصام يونس
- المؤلفون بالإنجليزي
- Mohammed Issam Younis
- الملخص العربي
- This paper reviews the state - of - the - art and the art - of - the - practice of the classification machine learning algorithms. In addition, th is paper proposes a novel input - output relation classification and testing strategy called Minimum Maximum Strategy (MMS). Internally, MMS derives the classification rules based on minimum - maxi mum values of attributes for each class till all entries in a data set are covered at least one. In doing so, MMS achieves 100% classification accuracy as well as mining the data set which facilitate building the classification model. Moreover, unlike othe r existing algorithm MMS generates instances for testing based on the boundary value analysis. As a proof of concept, MMS is used to build a classifier and test instances for the famous IRIS data set . Encouraging results are obtained from experimentations on the accuracy against well - known classification algorithms as well as the effectiveness of the test data generated by the MMS. Finally, it should be mentioned that all experiments are done using th e WEKA machine learning tool.
- الملخص الانجليزي
- This paper reviews the state - of - the - art and the art - of - the - practice of the classification machine learning algorithms. In addition, th is paper proposes a novel input - output relation classification and testing strategy called Minimum Maximum Strategy (MMS). Internally, MMS derives the classification rules based on minimum - maxi mum values of attributes for each class till all entries in a data set are covered at least one. In doing so, MMS achieves 100% classification accuracy as well as mining the data set which facilitate building the classification model. Moreover, unlike othe r existing algorithm MMS generates instances for testing based on the boundary value analysis. As a proof of concept, MMS is used to build a classifier and test instances for the famous IRIS data set . Encouraging results are obtained from experimentations on the accuracy against well - known classification algorithms as well as the effectiveness of the test data generated by the MMS. Finally, it should be mentioned that all experiments are done using th e WEKA machine learning tool.
- تاريخ النشر
- 01/08/2015
- الناشر
- International Journal of Computing Academic Research (IJCAR)
- رقم المجلد
- 4
- رقم العدد
- 4
- ISSN/ISBN
- 2305-9184
- الصفحات
- 162-178
- رابط الملف
- تحميل (491 مرات التحميل)
- رابط خارجي
- http://www.meacse.org/ijcar/archives/66.pdf
- الكلمات المفتاحية
- lassification ; Input - output relation ; machine learning ; data mining ; boundary value analysis ; evaluation metrics ; evaluation matrix ; learning ; testing.