Effect of Normazilation Techniques on Data Mining Methods: Case Study for Dairy Cattle Data

Aslı Akıllı; Hülya ATIL

Effect of Normazilation Techniques on Data Mining Methods: Case Study for Dairy Cattle Data

Kitapta bölümler > Mesleki Kitap · 2020

Yazarlar: Aslı Akıllı, Hülya ATIL

Veri Madenciliği Yöntemleri: Tarım alanında Uygulamaları, Mesleki Kitap, Rating Academy Yayınları, Tanınmış Ulusal, Ulusal, 2020, ss 127-147, Türkiye, Çanakkale

Özet

Abstract: Data mining methods are widely used in animal production areas as in many applied sciences. Data pre-processing and analytics of data set are very important in data mining analysis. Improvements in data mining performance are provided by using various normalization and standardization methods during the training phase. The aim of this study was to investigate the effect of eight normalization techniques “Z-Score, Min-Max, D-Min-Max, Median, Sigmoid, Decimal Scaling, Median and MAD, Tanh-Estimators” on artificial neural network, LSSVM and K-Means performance for this case in animal science. For this purpose, data sets were analyzed to perform clustering analysis based on milk yield characteristics. According to the result of analyses, most successful prediction performance is obtained using Bayesian Regularization algorithm with D-Min-Max normalization technique (R2Adj = 0.9997, RMSE= 0.0003, MAPE= 101.82 for test set. Results of analysis show that the data structures are effective in modeling, and the normalization process affects data mining efficiency.<o:p></o:p> Keywords: Agricultural data, back-propagation algorithms, data mining, K-Means, neural network, normalization, data pre-processing, data mining, least square support vector machines. <o:p></o:p>

Detay görüntüle