scholarly journals Estudio comparativo de técnicas de minería de datos para la predicción de rutas de huracanes

2021 ◽  
Vol 4 (1) ◽  
pp. 43-52
Author(s):  
M.A. Coronado Arjona ◽  
V. M. Bianchi Rosado ◽  
J. A. Vivas Burgos ◽  
M.A. Perera Collí
Keyword(s):  
Linear Regression ◽  
Multilayer Perceptron ◽  
Large Scale ◽  
Nearest Neighbors ◽  
K Nearest Neighbors ◽  
Weather Phenomenon ◽  
Prediction Technique ◽  
Linear Regression Technique ◽  
Hurricane Prediction ◽  
Prediction Techniques

Los huracanes son las tormentas más grandes y violentas que pueden existir sobre la tierra. Su peligrosidad radica en la velocidad que pueden alcanzar sus vientos, llegando a superar los 250 kilómetros por hora y desatando 9 billones de litros de lluvia al día, en consecuencia, sus efectos son a gran escala y con frecuencia muy destructivos en pérdidas humanas y materiales. A sabiendas de la inexactitud en las trayectorias, muchos habitantes esperan hasta el último momento antes de abandonar su hogar y pertenencias con la esperanza de que el fenómeno meteorológico cambie su curso. Es por esto que surge la necesidad de determinar la mejor técnica para predecir rutas de huracanes. El estudio consistió en entrenar los algoritmos de las técnicas de predicción, regresión lineal, k vecinos más cercanos y perceptrón multicapa, para obtener los modelos que permitan la comparación de datos predictivos con las trayectorias reales de huracanes y así determinar la exactitud de la predicción. Se encontró que la técnica de regresión lineal obtuvo los mejores resultados. Hurricanes are the largest and most violent storms that exist on Earth. Their dangerousness lies in the speed that can reach their winds, reaching over 250 km per hour and unleashing 9 billion liters of rain a day, so their effects are large scale and very destructive. Due to the effects mentioned before, the number of human and material losses are high. This is because of the inaccuracy in trajectories and many inhabitants wait until the last moment to leave their home and belongings, in the hope that the weather phenomenon will change its course. In this way arises the need to find the best hurricane prediction technique. The study consisted in training the algorithms of prediction techniques, linear regression, k nearest neighbors and multilayer perceptron, to obtain the models that allow the comparison of predictive data with the actual hurricane trajectories and thus determine the accuracy of the prediction. It was found that the linear regression technique obtained the best results.

Speeding up k-Nearest Neighbors classifier for large-scale multi-label learning on GPUs

2019 ◽  
Vol 354 ◽  
pp. 10-19 ◽  
Author(s):  
Przemysław Skryjomski ◽  
Bartosz Krawczyk ◽  
Alberto Cano
Keyword(s):  
Large Scale ◽  
Nearest Neighbors ◽  
K Nearest Neighbors

scholarly journals Software Development Effort Duration and Cost Estimation using Linear Regression and K-Nearest Neighbors Machine Learning Algorithms

2019 ◽  
Vol 9 (2) ◽  
pp. 1043-1047
Keyword(s):  
Machine Learning ◽  
Linear Regression ◽  
Software Development ◽  
Cost Estimation ◽  
Performance Metrics ◽  
Nearest Neighbors ◽  
Machine Learning Algorithms ◽  
Development Effort ◽  
Effort Estimation ◽  
K Nearest Neighbors

Effort estimation is a crucial step that leads to Duration estimation and cost estimation in software development. Estimations done in the initial stage of projects are based on requirements that may lead to success or failure of the project. Accurate estimations lead to success and inaccurate estimates lead to failure. There is no one particular method which cloud do accurate estimations. In this work, we propose Machine learning techniques linear regression and K-nearest Neighbors to predict Software Effort estimation using COCOMO81, COCOMONasa, and COCOMONasa2 datasets. The results obtained from these two methods have been compared. The 80% data in data sets used for training and remaining used as the test set. The correlation coefficient, Mean squared error (MSE) and Mean magnitude relative error (MMRE) are used as performance metrics. The experimental results show that these models forecast the software effort accurately.

scholarly journals Automated System for Buying products through IoT Commerce

2019 ◽  
Vol 8 (3) ◽  
pp. 255-257
Keyword(s):  
Linear Regression ◽  
Vital Role ◽  
Regression Technique ◽  
Automated System ◽  
The Internet ◽  
Internet Commerce ◽  
Network Problem ◽  
Linear Regression Technique ◽  
Prediction Techniques ◽  
Internet Of Thing

IoT commerce is the Internet of thing commerce in which buying and selling of products take place in automize way. IoT Commerce plays a vital role in the field of Internet commerce. IoT commerce is going to capture the market in 2022, and the whole world is becoming dependent on IoT commerce. So it is necessary to implement a method where prediction in IoT commerce takes place. In our research, we have implemented a method where buying and selling of products take place in an automated way. If server issue or network problem arises in this case by applying prediction techniques automatically product is delivered to the customer on time. The final result is the successful delivery of the product to the customer. For acquiring prediction results we have used linear regression technique. Where on the basis of previous data in the absence of network automatically order is placed on website and product will be delivered.

k ‐Nearest neighbors local linear regression for functional and missing data at random

2020 ◽  
Vol 75 (1) ◽  
pp. 42-65
Author(s):  
Mustapha Rachdi ◽  
Ali Laksaci ◽  
Zoulikha Kaid ◽  
Abbassia Benchiha ◽  
Fahimah A. Al‐Awadhi
Keyword(s):  
Missing Data ◽  
Linear Regression ◽  
Nearest Neighbors ◽  
Local Linear Regression ◽  
K Nearest Neighbors ◽  
Local Linear

scholarly journals Improving sub-pixel imperviousness change prediction by ensembling heterogeneous non-linear regression models

2016 ◽  
Vol 65 (2) ◽  
pp. 193-218 ◽  
Author(s):  
Wojciech Drzewiecki
Keyword(s):  
Random Forest ◽  
Linear Regression ◽  
Regression Models ◽  
Nearest Neighbors ◽  
Linear Regression Models ◽  
K Nearest Neighbors ◽  
Heterogeneous Model ◽  
Change Assessment ◽  
Non Linear ◽  
Model Ensembles

Abstract In this work nine non-linear regression models were compared for sub-pixel impervious surface area mapping from Landsat images. The comparison was done in three study areas both for accuracy of imperviousness coverage evaluation in individual points in time and accuracy of imperviousness change assessment. The performance of individual machine learning algorithms (Cubist, Random Forest, stochastic gradient boosting of regression trees, k-nearest neighbors regression, random k-nearest neighbors regression, Multivariate Adaptive Regression Splines, averaged neural networks, and support vector machines with polynomial and radial kernels) was also compared with the performance of heterogeneous model ensembles constructed from the best models trained using particular techniques. The results proved that in case of sub-pixel evaluation the most accurate prediction of change may not necessarily be based on the most accurate individual assessments. When single methods are considered, based on obtained results Cubist algorithm may be advised for Landsat based mapping of imperviousness for single dates. However, Random Forest may be endorsed when the most reliable evaluation of imperviousness change is the primary goal. It gave lower accuracies for individual assessments, but better prediction of change due to more correlated errors of individual predictions. Heterogeneous model ensembles performed for individual time points assessments at least as well as the best individual models. In case of imperviousness change assessment the ensembles always outperformed single model approaches. It means that it is possible to improve the accuracy of sub-pixel imperviousness change assessment using ensembles of heterogeneous non-linear regression models.

scholarly journals Utilização de Modelos Computacionais Baseados em Classificadores Para Predição da Dislexia em Crianças

2021 ◽  
Author(s):  
Ronieri Nogueira de Sousa ◽  
Roney Nogueira de Sousa ◽  
Rhyan Ximenes de Brito ◽  
Janaide Nogueira de Sousa Ximenes
Keyword(s):  
Support Vector Machine ◽  
Random Forest ◽  
Radial Basis Function ◽  
Basis Function ◽  
Multilayer Perceptron ◽  
Nearest Neighbors ◽  
Support Vector ◽  
Z Score ◽  
K Nearest Neighbors ◽  
Learning Machine

A dislexia é uma das dificuldades de aprendizagem mais comum nas salas de aula. Dessa forma o estudo teve como finalidade a classificação de crianças com ou sem dislexia através da aplicação de técnicas de Inteligência Computacional (IC). Para a metodologia utilizou-se de uma base de dados pública e da aplicação das arquiteturas neurais, Multilayer Perceptron (MLP), Radial Basis Function (RBF) e Extreme Learning Machine (ELM) e dos classificadores estatísticos, Support Vector Machine (SVM), Random Forest (RF) e K-Nearest Neighbors (K-NN), assim como das técnicas k-fold, SMOTE e normalização z-score. Os resultados demonstraram que o classificador SVM obteve a melhor taxa média de acerto com 98,03% de acurácia.

Sign in / Sign up

Export Citation Format

Share Document