Bu çalışmada çok değişkenli uyarlanabilir regresyon eğrileri (MARS) ve TreeNet gradyan arttırma makinesi (TreeNet) isimli regresyon tabanlı yöntemler kullanılarak çözünmüş oksijen (ÇO) konsantrasyonu modellemesi amaçlanmıştır. Modelleme çalışmasında kentsel atıksuları bünyesine alarak yer yer kirlenen Harşit Çayı (Gümüşhane) üzerinde belirlenmiş altı su kalitesi gözlem istasyonunda, 15 gün aralıklarla ve 24 kez yerinde gerçekleştirilen ÇO konsantrasyonu (mg/L), sıcaklık (°C), pH ve elektriksel iletkenlik (mS/cm) ölçümleri yanı sıra akarsudan alınan su örneklerinde laboratuvarda gerçekleştirilen sertlik (°dH) tayinleri neticesinde elde edilen veriler kullanılmıştır. Elde edilen veri setinin % 80’i kurulan modellerin eğitilmesinde geriye kalan % 20’si ise söz konusu modellerin test edilmesinde kullanılmıştır. Kurulan modellerin eğitim ve test veri seti performanslarını değerlendirmek amacıyla ortalama karesel hatanın karekökü (OKHK), ortalama mutlak hata (OMH), ortalama rölatif hata (ORH) ve determinasyon katsayısı (R2) performans istatistikleri kullanılmıştır. En düşük OKHK, OMH ve ORH ile en yüksek R2 değerleri eğitim veri seti için sırasıyla 0,2247 mg/L, 0,0666 mg/L, % 0,66 ve 0,9995 olarak TreeNet yönteminden, test veri seti için ise 0,2911 mg/L, 0,2336 mg/L, % 2,27 ve 0,9992 olarak MARS yönteminden elde edilmiştir. Her iki veri seti için ortalamalar dikkate alındığında ise, MARS yönteminden elde edilen performans değerlerinin TreeNet yönteminden elde edilenlere kıyasla daha iyi olduğu sonucuna ulaşılmıştır.
This study aimed to model the stream dissolved oxygen (DO) concentration using the multivariate adaptive regression splines (MARS) and TreeNet gradient boosting machine (TreeNet) methods. The water quality indicators employed for the modeling studies were the stream DO concentration (mg/L), temperature (°C), pH, and electrical conductivity (mS/cm), as well as hardness (°dH). These indicators were measured semimonthly during a year for six monitoring sites selected in untreated wastewater impacted urban stream, namely Harşit, Gümüşhane Province. The stream water quality data for each indicator were 144, 80 % for the training, and the rest for the testing. To evaluate the performance for the training and testing data sets of the models, four performance statistics, i.e., root mean square error (RMSE), mean absolute error (MAE), mean relative error (MRE), and coefficient of determination (R2), were computed. On the one hand, the TreeNet method provided better results for the training data set. On the other hand, the MARS method provided better results for the testing data set. The lowest RMSE, MAE, and MRE and highest R2 values were calculated as 0.2247 mg/L, 0.0666 mg/L, 0.66 %, and 0.9995 for the training data sets, and 0.2911 mg/L, 0.2336 mg/L, 2.27 %, and 0.9992 %, for the test data sets, respectively. It was concluded that the MARS method had a better performance than to the TreeNet method considering the mean values for both data sets.
Primary Language | Turkish |
---|---|
Subjects | Civil Engineering |
Journal Section | Research Articles |
Authors | |
Publication Date | April 30, 2022 |
Submission Date | June 10, 2021 |
Acceptance Date | February 23, 2022 |
Published in Issue | Year 2022 Volume: 27 Issue: 1 |
Announcements:
30.03.2021-Beginning with our April 2021 (26/1) issue, in accordance with the new criteria of TR-Dizin, the Declaration of Conflict of Interest and the Declaration of Author Contribution forms fulfilled and signed by all authors are required as well as the Copyright form during the initial submission of the manuscript. Furthermore two new sections, i.e. ‘Conflict of Interest’ and ‘Author Contribution’, should be added to the manuscript. Links of those forms that should be submitted with the initial manuscript can be found in our 'Author Guidelines' and 'Submission Procedure' pages. The manuscript template is also updated. For articles reviewed and accepted for publication in our 2021 and ongoing issues and for articles currently under review process, those forms should also be fulfilled, signed and uploaded to the system by authors.