Using the gradient boosting decision tree (GBDT) algorithm for a train delay prediction model considering the delay propagation feature

URN:NBN:SI:doc-WAO2J26W2021Li, K. H.Liao, L.Ma, W. G.Yu, Q.Zhang, Y. D.documents/doc/W/URN_NBN_SI_doc-WAO2J26W_001.pdfdocuments/doc/W/URN_NBN_SI_doc-WAO2J26W_001.txt163articlestr. 285-29610.14743/apem2021.3.4001854-6250270066947URN:NBN:SI:doc-WAO2J26WengFakulteta za strojništvo, Inštitut za proizvodno strojništvoAdvances in production engineering and managementBYactual train operation datadelay propagation feature extractiondelay type identificationdensity-based spatial clustering of applications with noise (DBSCAN)ekstrakcija značilnosti širjenja zakasnitvegradient boosting decision tree (GBDT)identifikacija tipa zakasnitvek-nearest neighbor (KNN)multilayer perceptron (MLP)naključni gozdnapoved zamud vlakovpodatki o dejanskem obratovanju vlakovprostorsko združevanje aplikacij s šumom na podlagi gostoterandom forest (RF)regresija podpornih vektorjevsupport vector regression (SVR)train delay predictionUsing the gradient boosting decision tree (GBDT) algorithm for a train delay prediction model considering the delay propagation feature