检索结果-内蒙古大学图书馆

An Insight into Machine Learning Algorithms to Map the Occurrence of the Soil Mattic Horizon in the Northeastern Qinghai-Tibetan Plateau

引用

Pedosphere 2018年第5期28卷 739-750页

作者： ZHI Junjun ZHANG Ganlin YANG Renmin YANG Fei JIN Chengwei LIU Feng SONG Xiaodong ZHAO Yuguo LI Decheng State Key Laboratory of Soil and Sustainable Agriculture Institute of Soil Science Chinese Academy of Sciences Nanjing 210008(China) University of Chinese Academy of Sciences Beijing 100049 (China)

Soil diagnostic horizons, which each have a set of quantified properties, play a key role in soil classification. However, they are difficult to predict, and few attempts have been made to map their spatial occurrence. We evaluated and compared four machine learning algorithms, namely, the classification and regression tree(CART), random forest(RF), boosted regression trees(BRT), and support vector machine(SVM), to map the occurrence of the soil mattic horizon in the northeastern Qinghai-Tibetan Plateau using readily available ancillary data. The mechanisms of resampling and ensemble techniques significantly improved prediction accuracies(measured based on area under the receiver operator characteristic curve score(AUC)) and produced more stable results for the BRT(AUC of 0.921 ± 0.012, mean ± standard deviation) and RF(0.908 ± 0.013) algorithms compared to the CART algorithm(0.784 ± 0.012), which is the most commonly used machine learning method. Although the SVM algorithm yielded a comparable AUC value(0.906 ± 0.006) to the RF and BRT algorithms, it is sensitive to parameter settings, which are extremely ***, we consider it inadequate for occurrence-distribution modeling. Considering the obvious advantages of high prediction accuracy, robustness to parameter settings, the ability to estimate uncertainty in prediction, and easy interpretation of predictor variables, BRT seems to be the most desirable method. These results provide an insight into the use of machine learning algorithms to map the mattic horizon and potentially other soil diagnostic horizons.

关键词： boosted regression trees classification and regression tree digital soil mapping random forest soil diagnostic horizons,support vector machine

来源：评论

学校读者我要写书评

暂无评论

Methane hydrate: Modeling and assessing the experimental data of incipient stability conditions

引用

JOURNAL OF DISPERSION SCIENCE AND TECHNOLOGY 2018年第6期39卷 848-861页

作者： Ghiasi, Mohammad M. Noorollahi, Younes Aslani, Alireza Univ Tehran Fac New Sci & Technol Dept Renewable Energies North Kargar St Tehran *** Iran

Accurate knowledge of equilibrium conditions for methane hydrates dissociation is crucial to implement an appropriate hydrate-based technology/process. Hence, much importance is attached to utilize techniques for the modeling of hydrate stability zones and the existence of reliable data as well. This research employs the published experimental data, from 1940 to 2016, for modeling the incipient stability conditions of methane hydrate in pure water using classification and regression tree (CART) as a novel methodology in gas hydrates research. Least squares support vector machine (LSSVM) and artificial neural network (ANN) methods were selected as a basis for comparison. Furthermore, the Leverage mathematical approach was used for evaluating the quality of the data as well as the correctness of the CART model. Results confirmed that the developed tree-based model provides excellent outcomes and no model can rival it for accuracy. In addition, applying the Leverage algorithm specified that: (a) the CART model is statistically valid and correct;(b) the published experimental data in the literature are different in quality;(c) there is no doubtful data in the studied database;and (d) the published data in the temperature range from 148.8 to 238.8K are not sufficient compared with the whole of the collected database. [GRAPHICS] .

关键词： classification and regression tree clathrate methane hydrate modeling outlier detection

来源：评论

学校读者我要写书评

暂无评论

Taxonomy of influential factors for predicting pollutant first flush in urban stormwater runoff

引用

WATER RESEARCH 2019年 166卷 115075-000页

作者： Perera, Thamali McGree, James Egodawatta, Prasanna Jinadasa, K. B. S. N. Goonetilleke, Ashantha Queensland Univ Technol Sci & Engn Fac GPO Box 2434 Brisbane Qld 4001 Australia Univ Srijayewardenepura Dept Math Nugegoda 10250 Sri Lanka Univ Peradeniya Dept Civil Engn Peradeniya 20400 Sri Lanka

Pollutant first flush in urban stormwater runoff is an important phenomenon influenced by a range of rainfall and catchment related variables. Even though numerous studies have been undertaken to mathematically define the first flush and the influential variables of first flush, limited research have been carried out to rank such variables in terms of their level of importance in generating first flush. Identifying the degree of importance of the variables is critical for accurate predictions of first flush occurrence and understanding the main drivers of first flush. This research study undertook a comprehensive analysis of the variables influencing the predictions of first flush occurrence and their relative importance. The study results are expected to contribute to more accurate predictions of first flush by affording greater importance to the highly ranked factors and their impacts. The study outcomes confirmed that total rainfall depth was the most important variable influencing the prediction of first flush events while the maximum intensity was the second. Rain duration, runoff depth, runoff peak and average intensity were the next four most important variables. Antecedent dry period and effective impervious area fraction had relatively low ranking while the time of concentration and the event mean concentration were found to be the least important variables. Furthermore, the study outcomes highlight that the use of a combination of variables and due consideration of their interactions can yield better results than considering their individual roles. (C) 2019 Elsevier Ltd. All rights reserved.

关键词： Pollutant first flush Stormwater treatment design classification and regression tree Random forest Stormwater quality Stormwater pollutant processes

来源：评论

学校读者我要写书评

暂无评论

Developing sarcopenia criteria and cutoffs for an older Caucasian cohort - a strictly biometrical approach

引用

CLINICAL INTERVENTIONS IN AGING 2018年 13卷 1365-1373页

作者： Kemmler, Wolfgang von Stengel, Simon Kohl, Matthias Univ Erlangen Nurnberg Inst Med Phys Henkestr 91 D-91052 Erlangen Germany Univ Furtwangen Dept Med & Life Sci Villingen Schwenningen Germany

Background: The aim of this strictly statistical approach was to provide a figure discrimination in a homogeneous cohort that is based on a main component, which includes disability, physical performance, and autonomy parameters. Methods: We used data of 939 community-dwelling men aged >= 70 years, living in the area of Erlangen-Nurnberg, Germany. Briefly, we conducted a scaled principal component analysis based on criteria related to "physical function", "disability", "weakness", and "autonomy" to identify men who are likely to have sarcopenia as per the recognized sarcopenia criteria. Next, we applied fast-and-frugal decision trees, logistic regression, and classification and regression decision trees to classify men with and without sarcopenia, applying the 5% prevalence rate identified for this cohort by recent studies. Results: In summary, the best fast-and-frugal decision trees included gait velocity, handgrip strength, and two skeletal muscle mass indices (SMI)-appendicular skeletal muscle mass (ASMM)/body mass index (BMI) and ASMM/height(2). Briefly, men below the cutoff point of 1.012 m/s for gait velocity were directly classified as sarcopenic. Faster men with a handgrip strength of >34.5 kg were excluded from further screening, while their weaker peers were assessed for SMI. Firstly, an ASMM/BMI-based exclusion criterion of >0.886 indicates no sarcopenia;while in men with a lower BMI-based SMI, an ASMM/height(2) of <7.25 kg/m(2) indicates sarcopenia. Of importance, about 72% of the participants can be classified without an SMI assessment. Conclusion: The present approach that applied recognized sarcopenia criteria and was based on a predominately functional understanding of sarcopenia provided a simple and feasible decision rule for sarcopenia discrimination. In summary, we consider our approach as a strictly biometrical contribution within the development of sarcopenia screening methods. However, our tool needs to be further evaluated to validate its app

关键词： sarcopenia screening sarcopenia cutoff points classification and regression tree fast-and-frugal decision trees Caucasian men aged 70+

来源：评论

学校读者我要写书评

暂无评论

Predicting interstate motor carrier crash rate level using classification models

引用

ACCIDENT ANALYSIS AND PREVENTION 2018年 120卷 211-218页

作者： Liu, Jundi Boyle, Linda N. Banerjee, Ashis G. Univ Washington Ind & Syst Engn Seattle WA 98195 USA Univ Washington Civil & Environm Engn Seattle WA 98195 USA Univ Washington Mech Engn Seattle WA 98195 USA

Ensuring safe operations of large commercial vehicles (motor carriers) remains an important challenge, particularly in the United States. While the federal regulatory agency has instituted a compliance review-based rating method to encourage carriers to improve their safety levels, concerns have been expressed regarding the effectiveness of the current ratings. In this paper, we consider a crash rate level (high, medium, and low) rather than a compliance review-based rating (satisfactory, conditional satisfactory, and unsatisfactory). We demonstrate an automated way of predicting the crash rate levels for each carrier using three different classification models (Artificial Neural Network, classification and regression tree (CART), and Support Vector Machine) and three separate variable selection methods (Empirical Evidence, Multiple Factor Analysis, Garson's algorithm). The predicted crash rate levels (high, low) are compared to the assigned levels based on the current safety rating method. The results indicate the feasibility of crash rate level as an effective measure of carrier safety, with CART having the best performance.

关键词： Motor carrier safety Crash rate Variable selection classification and regression tree

来源：评论

学校读者我要写书评

暂无评论

Using handwriting to infer a writer's country of origin for forensic intelligence purposes

引用

FORENSIC SCIENCE INTERNATIONAL 2018年 282卷 144-156页

作者： Agius, Anna Morelato, Marie Moret, Sebastien Chadwick, Scott Jones, Kylie Epple, Rochelle Brown, James Roux, Claude Univ Technol Sydney Ctr Forens Sci POB 123 Broadway 2007 Australia Australian Fed Police Forens Specialist Operat GPO Box 401 Canberra ACT Australia Univ Technol Sydney Sch Math & Phys Sci POB 123 Broadway 2007 Australia

Forensic science has traditionally focused the majority of its resources and objectives towards addressing Court-related questions. However, this view restricts the contribution of forensic science to one process and results in a loss of information as the investigative and intelligence roles are largely neglected. A forensic science discipline suffering from this imbalance is handwriting examination, which may be characterised as a time consuming and subjective process that is mostly carried out towards the end of the investigation for the purpose of judicial proceedings. Individual and habitual characteristics are the major handwriting features exploited, however alternate information concerning the author's native language could potentially be used as a key element in an intelligence framework. This research focussed on the detection of characteristics that differentiate Vietnamese and English Australian writers based on their English handwriting. The study began with the extraction of handwriting characteristics from the writing of people from the two populations. The data was analysed using a logistic regression model and a classification and regression tree (CRT). Each recognised four class characteristics that were capable of distinguishing between the two nationalities. The logistic regression and CRT models were both capable of correctly predicting 93% of cases. Their predictive capabilities were then tested and supported using blind exemplars in order to mirror casework settings. It appeared that when using their respective class characteristics, the two models were capable of differentiating English Australians from Vietnamese in the data set. This proof of concept research demonstrated the plausibility of exploiting this additional information from a handwriting trace and taking advantage of it in an intelligence-led framework. (c) 2017 Elsevier B.V. All rights reserved.

关键词： Documents Writers Intelligence Logistic regression model classification and regression tree

来源：评论

学校读者我要写书评

暂无评论

Estimation of Forest Crown Density Using Pleiades Satellite Data and Nonparametric classification Method

引用

JOURNAL OF THE INDIAN SOCIETY OF REMOTE SENSING 2018年第7期46卷 1151-1158页

作者： Kalbi, Siavash Hassanvand, Mohammad Nabi Soosani, Javad Abrary, Kambiz Naghavi, Hamed Sari Agr & Nat Resource Sari Mazandaran Iran Lorestan Univ Khorramabad Lorestan Iran

The Zagros forests in western Iran are a crucial source of environmental services, but are severely threatened by climatic and anthropological constraints. One crucial forest parameter is the amount of canopy coverage which enables an indirect assessment of aboveground biomass and, in turn, the carbon emission and sequestration. The aim of this study is estimation of forest crown density using Pleiades satellite data and three non-parametric classification include random forest (RF), boosted regression tree (BRT) and classification and regression tree (CART) in Kaka Reza of Lorestan province, western Iran. For this purpose, we then ensured the accuracy of geometric images, creating the necessary processing such as vegetation indices, principal component analysis and texture analysis was performed on the original bands using a random-systematic sampling design, 96 sample plots were taken. Results showed that RF compared to the two other algorithms with overall accuracy of 75% and kappa coefficient of 0.73 could better classify the forest crown density, while the CART method had the lowest accuracy with overall accuracy of 71% and kappa coefficient of 0.68. Also results showed BRT had overall accuracy of 71% and kappa coefficient of 0.68. Overall results showed Pleiades satellite data and non-parametric classification method had high capability for separation crown density in Zagros region.

关键词： Kaka Reza Random forest Boosted regression tree classification and regression tree Texture analyse

来源：评论

学校读者我要写书评

暂无评论

A Comparison of Logistic regression, classification and regression tree, and Neural Networks Models in Predicting Violent Re-Offending

引用

JOURNAL OF QUANTITATIVE CRIMINOLOGY 2011年第4期27卷 547-573页

作者： Liu, Yuan Y. Yang, Min Ramsay, Malcolm Li, Xiao S. Coid, Jeremy W. Sichuan Univ Sch Publ Hlth Dept Hlth Stat Chengdu 610041 Peoples R China Univ Nottingham Sch Community Hlth Sci Div Psychiat Nottingham NG7 2TU England Minist Justice Partnerships & Hlth Strategy Unit London SW1H 9AJ England Queen Mary Univ London Forens Psychiat Res Unit London EC1A 7BE England

Previous studies that have compared logistic regression (LR), classification and regression tree (CART), and neural networks (NNs) models for their predictive validity have shown inconsistent results in demonstrating superiority of any one model. The three models were tested in a prospective sample of 1225 UK male prisoners followed up for a mean of 3.31 years after release. Items in a widely-used risk assessment instrument (the Historical, Clinical, Risk Management-20, or HCR-20) were used as predictors and violent reconvictions as outcome. Multi-validation procedure was used to reduce sampling error in reporting the predictive accuracy. The low base rate was controlled by using different measures in the three models to minimize prediction error and achieve a more balanced classification. Overall accuracy of the three models varied between 0.59 and 0.67, with an overall AUC range of 0.65-0.72. Although the performance of NNs was slightly better than that of LR and CART models, it did not demonstrate a significant improvement.

关键词： Violence reconviction Risk assessment Neural networks classification and regression tree HCR-20

来源：评论

学校读者我要写书评

暂无评论

Enhancing prognosis prediction using pre-treatment nodal SUVmax and HPV status in cervical squamous cell carcinoma

引用

CANCER IMAGING 2019年第1期19卷 43-43页

作者： Hong, Chae Moon Park, Shin-Hyung Chong, Gun Oh Lee, Yoon Hee Jeong, Ju Hye Lee, Sang-Woo Lee, Jaetae Ahn, Byeong-Cheol Jeong, Shin Young Kyungpook Natl Univ Sch Med Dept Nucl Med Daegu South Korea Kyungpook Natl Univ Sch Med Dept Obstet & Gynecol Daegu South Korea Kyungpook Natl Univ Sch Med Dept Radiat Oncol Daegu South Korea Kyungpook Natl Univ Hosp Dept Nucl Med 130 Dongdeok Ro Daegu 41944 South Korea Kyungpook Natl Univ Chilgok Hosp Sch Med Dept Nucl Med 807 Hoguk Ro Daegu 41404 South Korea Kyungpook Natl Univ Chilgok Hosp Sch Med Dept Obstet & Gynecol 807 Hoguk Ro Daegu 41404 South Korea Kyungpook Natl Univ Chilgok Hosp Dept Radiat Oncol 807 Hoguk Ro Daegu 41404 South Korea

BackgroundThis study was to evaluate the prognostic value of metabolic parameters on F-18-FDG PET/CT and the status of human papillomavirus (HPV) infection and known prognostic variables for predicting tumor recurrence and investigating a prognostic model in patients with locally advanced cervical cancer treated with concurrent chemoradiotherapy (CCRT).MethodsA total of 129 patients with cervical squamous cell carcinoma who underwent initial CCRT were eligible for this study. Univariate and multivariate analyses were performed using traditional prognostic factors, metabolic parameters, and HPV infection. classification and regression decision tree (CART) was used to establish new *** 129 patients, 29 patients (22.5%) had recurrence after a median follow-up of 60months (range, 3-125months). Tumor size, para-aortic lymph node metastasis, nodal SUVmax, and HPV infection status were identified as independent prognostic factors by multivariate analysis. The CART analysis classified the patients into three groups. The first node was nodal SUVmax, and HPV status was the second node for patients with nodal SUVmax 7.49;Group A (nodal SUVmax 7.49 and HPV positive, HR 1.0), Group B (nodal SUVmax 7.49 and HPV negative, HR 3.56), and Group C (nodal SUVmax >7.49, HR 10.13). Disease-free survival was significantly different among the three groups (p<0.001).ConclusionThe nodal SUVmax on F-18 FDG PET/CT and HPV infection status before CCRT are powerful independent prognostic factors for the prediction of disease-free survival in patients with cervical squamous cell carcinoma who underwent initial CCRT. We also suggest a simple prognosis prediction model using pre-treatment FDG PET/CT and HPV genotyping;however, it needs further validation in an independent dataset.

关键词： Cervical cancer FDG PET CT Lymph node Human papilloma virus Prognosis prediction classification and regression tree Concurrent chemoradiotherapy

来源：评论

学校读者我要写书评

暂无评论

Impact of initial coagulation and fibrinolytic markers on mortality in patients with severe blunt trauma: a multicentre retrospective observational study

引用

SCANDINAVIAN JOURNAL OF TRAUMA RESUSCITATION & EMERGENCY MEDICINE 2019年第1期27卷 25-25页

作者： Ishii, Kenta Kinoshita, Takahiro Kiridume, Kazutaka Watanabe, Atsushi Yamakawa, Kazuma Nakao, Shota Fujimi, Satoshi Matsuoka, Tetsuya Senshu Trauma & Crit Care Ctr Rinku Gen Med Ctr Dept Trauma & Crit Care 2-23 Rinku Orai Kita Osaka 5988577 Japan Osaka Gen Med Ctr Div Trauma & Surg Crit Care Sumiyoshi Ku 3-1-56 Bandai Higashi Osaka 5588558 Japan

BackgroundAcute coagulopathy is a well-known predictor of poor outcomes in patients with severe trauma. However, using coagulation and fibrinolytic markers, how one can best predict mortality to find out potential candidates for treatment of coagulopathy remains unclear. This study aimed to determine preferential markers and their optimal cut-off values for mortality *** conducted a retrospective observational study of patients with severe blunt trauma (injury severity score16) transferred directly from the scene to emergency departments at two trauma centres in Japan from January 2013 to December 2015. We investigated the impact and optimal cut-off values of initial coagulation (platelet counts, fibrinogen and prothrombin time-international normalised ratio) and a fibrinolytic marker (D-dimer) on 28-day mortality via classification and regression tree (CART) analysis. Multivariate logistic regression analysis confirmed the importance of these markers. Receiver operating characteristic curve analyses were used to examine the prediction accuracy for *** 666 patients with severe blunt trauma were analysed. CART analysis revealed that the initial discriminator was fibrinogen (cut-off, 130mg/dL) and the second discriminator was D-dimer (cut-off, 110g/mL in the lower fibrinogen subgroup;118g/mL in the higher fibrinogen subgroup). The 28-day mortality was 90.0% (lower fibrinogen, higher D-dimer), 27.8% (lower fibrinogen, lower D-dimer), 27.7% (higher fibrinogen, higher D-dimer) and 3.4% (higher fibrinogen, lower D-dimer). Multivariate logistic regression demonstrated that fibrinogen levels <130mg/dL (adjusted odds ratio [aOR], 9.55;95% confidence interval [CI], 4.50-22.60) and D-dimer 110g/mL (aOR, 5.89;95% CI, 2.78-12.70) were independently associated with 28-day mortality after adjusting for probability of survival by the trauma and injury severity score (TRISS Ps). Compared with the TRISS Ps alone (0.900;95% CI, 0.870-0.931), TRISS

关键词： Coagulopathy D-dimer Fibrinogen classification and regression tree Logistic regression Receiver operating characteristic curve Prediction model

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：