Wyniki wyszukiwania - Biblioteka Nauki

Ten serwis zostanie wyłączony 2025-02-11.

Nowa wersja platformy, zawierająca wyłącznie zasoby pełnotekstowe, jest już dostępna.
Przejdź na https://bibliotekanauki.pl

Ograniczanie wyników

Znaleziono wyników: 3

Liczba wyników na stronie

Wyniki wyszukiwania

Wyszukiwano:
w słowach kluczowych: tree-based models

Sortuj według:

Ogranicz wyniki do:

Gradient Boosting in Regression

100%

Gatnar E. , University of Economics in Katowice I. o. S.

tom 194

Szeroko stosowane w praktyce metody nieparametryczne wykorzystujące tzw. drzewa regresyjne mają jedną istotną wadę. Otóż wykazują one niestabilność, która oznacza, że niewielka zmiana wartości cech obiektów w zbiorze uczącym może prowadzić do powstania zupełnie innego modelu. Oczywiście wpływa to negatywnie na ich trafność prognostyczną. Tę wadę można jednak wyeliminować, dokonując agregacji kilku indywidualnych modeli w jeden. Znane są trzy metody agregacji modeli i wszystkie opierają się na losowaniu ze zwracaniem obiektów ze zbioru uczącego do kolejnych prób uczących: agregacja bootstrapowa (boosting), losowanie adaptacyjne (bagging) oraz metoda hybrydowa, łącząca elementy obu poprzednich. W analizie regresji szczególnie warto zastosować gradientową, sekwencyjną, odmianę metody boosting. W istocie polega ona wykorzystaniu drzew regrcsyjnych w kolejnych krokach do modelowania reszt dla modelu uzyskanego w poprzednim kroku.

The successful tree-based methodology has one serious disadvantage: lack of stability. That is, regression tree model depends on the training set and even small change in a predictor value could lead to a quite different model. In order to solve this problem single trees are combined into one model. There are three aggregation methods used in classification: bootstrap aggregation (bagging), adaptive resample and combine (boosting) and adaptive bagging (hybrid bagging-boosting procedure). In the field of regression a variant of boosting, i.e. gradient boosting, can be used. Friedman (1999) proved that boosting is equivalent to a stepwise function approximation in which in each step a regression tree models residuals from last step model.

Feature Selection and Multiple Model Approach in Discriminant Analysis

100%

Gatnar E. , Katowice University of Economics I. o. S.

tom 206

Significant improvement of model stability and prediction accuracy in classification and regression can be obtained by using the multiple model approach. In classification multiple models are built on the basis of training subsets (selected from the training set) and combined into an ensemble or a committee. Then the component models (classification trees) determine the predicted class by voting. In this paper some problems of feature selection for ensembles will be discussed. We propose a new correlation-based feature selection method combined with the wrapper approach.

Groundwater spring potential prediction using a deep-learning algorithm

63%

Moughani S. K. , Osmani A. , Nohani E. , Khoshtinat S. , Jalilian T. , Askari Z. , Heddam S. , Tiefenbacher J. P. , Hatamiafkoueieh J.

tom Vol. 72, no. 2

1033--1054

Information about water resources is crucial for sustainable development, and this issue is considered to be one of the most important concerns worldwide due to rapid industrialization and population growth. Countries in the semiarid region of the western Asia, like Iran, are dependent on groundwater resources so access to these resources is vital. This study maps surface spring potential on the Nourabad-Koohdasht Plain of Iran using a deep-learning algorithm called convolutional neural network (CNN), and the result was compared to predictions made with five advanced data-mining models: logistic model tree (LMT), LMT hybridized with bagging (BA-LMT), LMT hybridized with dagging (DA-LMT), LMT hybridized with random subspace (RS-LMT), and LMT hybridized with AdaBoost (AB-LMT). Frequency ratio was used to assess the strengths of relationships of each subclass layer to groundwater presence and evidential belief function revealed their effects on model uncertainty. The locations of 2463 springs were determined and showed that the northern part of the plain has the highest groundwater potential based on the density of springs. The data representing each of the spring locations were used for prediction modeling. Receiver operating characteristic (ROC) and area under the ROC curve (AUC) were used to evaluate the strengths of the predictions produced by the models. The results show that CNN (AUC = 0.885) provided the best prediction of spring locations. AB-LMT (AUC = 0.877) was second best, and BA-LMT (AUC = 0.876), DA-LMT (AUC = 0.856), RS-LMT (AUC = 0.846), and the standalone LMT model (AUC = 0.827) followed in rank. It can be concluded that the hybrid LMT models increased the predictive strength of the standalone LMT model when used to predict spring locations. These hybrid modeling methods may be used to improve sustainable groundwater management in the study region and in other regions as well.