Migratory Logistic Regression for Learning Concept Drift Between Two Data Sets With Application to UXO Sensing

Liao, Xiaozhong; Carin, Lawrence

doi:10.1109/tgrs.2008.2005268

Cited by 35 publications

(4 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Among the most popular approaches in natural hazard modelling, we have: frequency ratio (FR) Samanta et al, 2018), analytical hierarchy process (AHP) (Yalcin 2008;Stefanidis and Stathis, 2013), fuzzy logic (Perera and Lahat, 2014), logistic regression (LR) (Pradhan, 2010;Tehrany et al, 2014), arti cial neural networks (ANN) (Kia et al 2012;Lohani et al al., 2012) and weights of evidence (WoE) (Dahal et al 2008). Among all these approaches, the FR could be considered one of the simplest and most effective in ooding risk area mapping (Liao and Carin, 2009). It is a relatively new tool widely used for risk areas mapping several other complex natural disasters, such as landslides (Rahmati et al, 2016).…”

Section: Introductionmentioning

confidence: 99%

Flood susceptibility mapping in the Tongo Bassa watershed through GIS, remote sensing and frequency ratio model

Ebodé

2024

Preprint

View full text Add to dashboard Cite

Flooding constitutes a major problem for the inhabitants of Douala City in general and those of the Tongo Bassa watershed (TBW) in particular. Faced with this situation, public authorities need to put in place measures to mitigate the vulnerability of populations to these disasters. This article aims to map flooding risk areas in the TBW using geographic information system (GIS), field data (historical flood points), remote sensing data (Sentinel II image) and frequency ratio (FR) model. Six independent variables linked to floods were considered (land use and land cover patterns, altitudes, slopes, drainage densities, distances from watercourses and soil types). The map of land use patterns was obtained following the supervised classification of a Landsat 8 image using ENVI 5.2 software. The other five (05) variables were automatically generated from a digital terrain model using ArcGIS 10.2 software. The map produced shows that 1.41%, 8.88%, 28.51%, 33.86% and 27.33% of the basin area are respectively delimited into very low, low, medium, high and very high flood vulnerability classes. High and very high flooding risk areas (those where flooding is most likely to occur) occupy more than half of the basin (61.19%). These areas are characterized by significant imperviousness, low altitudes, weak slopes, significant proximity to watercourses and clayey soils. Most of the houses in the basin (66.92%) are located in areas affected by these two levels of exposure (high and very high). With respective success and prediction accuracy rates of 89% and 96.78%, a certain confidence deserves to be placed on the map of flooding risk areas produced. The results of this study could serve as decision-support tools in the context of the various developments undertaken by the public authorities, to fight against flooding in this basin.

show abstract

Section: Introductionmentioning

confidence: 99%

Flood susceptibility mapping in the Tongo Bassa watershed through GIS, remote sensing and frequency ratio model

Ebodé

2024

Preprint

View full text Add to dashboard Cite

show abstract

“…Remote sensing techniques [16,48,49] have proven to be an accurate method that can be easily applied in various regions across the globe [16,50,51]. The ratio (FR) method is an effective statistical technique with a simple yet comprehensible concept [52]. It enables base scenario analysis (BSA) to study the impact of different factors on flooding occurrences [13,24].…”

Section: Introductionmentioning

confidence: 99%

Flood Susceptibility Assessment through Statistical Models and HEC-RAS Analysis for Sustainable Management in Essaouira Province, Morocco

Khouz,

Trindade,

Santos

et al. 2023

Geosciences

View full text Add to dashboard Cite

Floods are natural disasters that often impact communities living in low-lying areas in the northern and central parts of Morocco. In this study, our aim was to create a flood susceptibility map using three methods; the hierarchy process (AHP) frequency ratio model (FR) and the weights of evidence (WoE) model. We extensively examined the area identified by these approaches using a hydraulic analysis software called HEC-RAS (version 6.3.1). Our analysis focused on the Essaouira watersheds in Morocco, where we identified around 197 flood locations. Out of these, we randomly selected 70% for modeling purposes while the remaining 30% were used for validation. Ten factors that influence floods were considered, such as slope, elevation, proximity to rivers, drainage density, stream order, land use patterns, rainfall data, lithology (permeability level) index (TWI), and curvature. We obtained these factors from data sources. Finally, we generated a flood susceptibility map and evaluated its accuracy by calculating the area under the curve (AUC). The validation results confirmed that all three models were robust and effective with an AUC of 90. Moreover, the research uncovered a trend of vulnerability with the most susceptible area being in close proximity to the city of Essaouira along the Oued Ksob. A detailed analysis using HEC-RAS was conducted at this identified location, pinpointing the village of Diabat as highly exposed. These findings hold significance for flood management, empowering decision makers, scholars, and urban planners to make informed choices and implement strategies that can minimize the impact of floods in susceptible regions while minimizing potential damages.

show abstract

“…For example, results of the MCDM model are subject to disparities as a result of biased expert judgment (Ghorbanzadeh et al, 2019;Paquette and Lowry, 2012). Moreover, results of the statistical models depend heavily on the sample size (Liao and Carin, 2009).On the other hand, hydrodynamic models convert discharge flows into flood depths or flood velocity (Al-Mulali et al, 2015). Although this model yields relatively accurate results for small basin, it is challenging to apply this model to entire area (Guo et al, 2012;van Emmerik et al, 2015).…”

Section: Introductionmentioning

confidence: 99%

How do multiple kernel functions in machine learning algorithms improve precision in flood probability mapping?

et al. 2022

View full text Add to dashboard Cite

With climate change, hydro-climatic hazards, i.e., floods in the Himalayas regions, are expected to worsen, thus, likely to affect humans and socio-economic growth. Precisely, the Koshi River basin (KRB) is often impacted by flooding over the year. However, studies on estimating and predicting floods still lack in this basin.This study aims at developing flood probability map using machine learning algorithms (MLAs): gaussian process regression (GPR) and support vector machine (SVM) with multiple kernel functions including Pearson VII function kernel (PUK), polynomial, normalized poly kernel, and radial basis kernel function (RBF). Historical flood locations with available topography, hydrogeology, and environmental datasets were further considered to build flood model.Two datasets were carefully chosen to measure the feasibility and robustness of MLAs: training dataset (location of floods between 2010 and 2019) and testing dataset (flood locations of 2020) with thirteen flood influencing factors. 2The validation of the MLAs was evaluated using a validation dataset and statistical indices such as the coefficient of determination (r 2 : 0.546~0.995), mean absolute error (MAE: 0.009~0.373), root mean square error (RMSE: 0.051~0.466), relative absolute error (RAE: 1.81~88.55%), and root-relative square error (RRSE: 10.19~91.00%).Results showed that the SVM-Pearson VII kernel (PUK) yielded better prediction than other algorithms. The resultant map from SVM-PUK revealed that 27.99% area with low, 39.91% area with medium, 31.00% with high, and 1.10% area with very high probabilities of flooding in the study area. The final flood probability map could add a greatt value to the effort of flood risk mitigation and planning processes in KRB.

show abstract

Migratory Logistic Regression for Learning Concept Drift Between Two Data Sets With Application to UXO Sensing

Abstract: Abstract-To achieve good generalization in supervised learning, the training and testing examples are usually required to be drawn from the same source distribution. In this paper we propose a method to relax this requirement in the context of logistic regression.

Cited by 35 publications

References 25 publications

Flood susceptibility mapping in the Tongo Bassa watershed through GIS, remote sensing and frequency ratio model

Flood susceptibility mapping in the Tongo Bassa watershed through GIS, remote sensing and frequency ratio model

Flood Susceptibility Assessment through Statistical Models and HEC-RAS Analysis for Sustainable Management in Essaouira Province, Morocco

How do multiple kernel functions in machine learning algorithms improve precision in flood probability mapping?

Contact Info

Product

Resources

About