Handling mixed categorical and numerical features in machine learning techniques /
Dina Tantawy Hassan Saleh
Handling mixed categorical and numerical features in machine learning techniques / معالجة الخصائص الرقمية والفئوية المختلطة في طرق تعليم الآلة Dina Tantawy Hassan Saleh ; Supervised Amir Fouad Sorial Atiya , Olfat Gamil Shaker - Cairo : Dina Tantawy Hassan Saleh , 2016 - 94 P. : charts , facsimiles ; 30cm
Thesis (M.Sc.) - Cairo University - Faculty of Engineering - Department of Computer Engineering
Most classiers cannot handle mixed categorical and numerical features directly. Each classier is usually built for only one type of features either categorical or numerical ones. This study analyzes the handling technique using mid-sized data on several dierent classiers. It also introduce a number of new techniques like dynamic simplex, catRank, Joint measure and devise the proof for kernelized versions of IOF, OF, Goodall, Lin and overlap measures. A case study on breast cancer prediction is also provided with some recommendation for both medical and machine learning parts. The study reveals an association between classiers and techniques and provides some recommendation for handling techniques usage
Classication Machine Learning Mixed Categorical and Numerical
Handling mixed categorical and numerical features in machine learning techniques / معالجة الخصائص الرقمية والفئوية المختلطة في طرق تعليم الآلة Dina Tantawy Hassan Saleh ; Supervised Amir Fouad Sorial Atiya , Olfat Gamil Shaker - Cairo : Dina Tantawy Hassan Saleh , 2016 - 94 P. : charts , facsimiles ; 30cm
Thesis (M.Sc.) - Cairo University - Faculty of Engineering - Department of Computer Engineering
Most classiers cannot handle mixed categorical and numerical features directly. Each classier is usually built for only one type of features either categorical or numerical ones. This study analyzes the handling technique using mid-sized data on several dierent classiers. It also introduce a number of new techniques like dynamic simplex, catRank, Joint measure and devise the proof for kernelized versions of IOF, OF, Goodall, Lin and overlap measures. A case study on breast cancer prediction is also provided with some recommendation for both medical and machine learning parts. The study reveals an association between classiers and techniques and provides some recommendation for handling techniques usage
Classication Machine Learning Mixed Categorical and Numerical