• 대한원격탐사학회지
• /
• 제36권6_3호
• /
• pp.1711-1720
• /
• 2020

최근 급속한 산업화와 도시화로 인해 인위적으로 발생하는 미세먼지(Particulate matter, PM)는 기상 조건에 따라 이동 및 분산되면서 피부와 호흡기 등 인체에 악영향을 미친다. 본 연구는 기상인자를 multiple linear regression(MLR), support vector machine(SVM), 그리고 random forest(RF) 모델의 입력자료로 하여 서울시 PM10 농도를 예측하고, 모델 간 성능을 비교 평가하는데 그 목적을 둔다. 먼저 서울시에 소재한 39개소 대기오염측정망(air quality monitoring sites, AQMS)에서 관측된 PM10 농도 자료를 8:2 비율로 구분하여 모델 훈련과 검증 데이터셋으로 사용되었다. 또한 기상관측소(automatic weather system, AWS)에서 관측되고 있는 자료 중 9개 기상인자(평균기온, 최고기온, 최저기온, 일 강수량, 평균풍속, 최대순간풍속, 최대순간풍속풍향, 황사발생유무, 상대습도)가 모델의 입력자료로 선정되었다. 각 AQMS에서 관측된 PM10 농도와 MLR, SVM, 그리고 RF 모델에 의해 예측된 PM10 농도 간 결정계수(R2)는 각각 0.260, 0.772, 그리고 0.793이었고, RF 모델이 PM10 농도 예측에 가장 높은 성능을 나타냈다. 특히 모델 검증에 사용되는 AQMS 중 관악구와 강남대로 AQMS는 상대적으로 AWS에 가까워 SVM과 RF 모델에서 높은 정확도를 나타냈다. 종로구 AQMS는 AWS에서 비교적 멀리 떨어져 있지만, 인접한 두 AQMS 데이터가 모델 학습에 사용되었기 때문에 두 모델에서 높은 정확도를 나타냈다. 반면 용산구 AQMS는 AQMS 및 AWS에서 비교적 멀리 떨어져 있기에 두 모델의 성능이 낮게 나타냈다.

• Bulletin of the Korean Chemical Society
• /
• 제33권2호
• /
• pp.613-619
• /
• 2012

We applied several machine learning methods for developing QSAR models for prediction of acute toxicity to fathead minnow. The multiple linear regression (MLR) and artificial neural network (ANN) method were applied to predict 96 h $LC_{50}$ (median lethal concentration) of 555 chemical compounds. Molecular descriptors based on 2D chemical structure were calculated by PreADMET program. The recursive partitioning (RP) model was used for grouping of mode of actions as reactive or narcosis, followed by MLR method of chemicals within the same mode of action. The MLR, ANN, and two RP-MLR models possessed correlation coefficients ($R^2$) as 0.553, 0.618, 0.632, and 0.605 on test set, respectively. The consensus model of ANN and two RP-MLR models was used as the best model on training set and showed good predictivity ($R^2$=0.663) on the test set.

• Advances in Computational Design
• /
• 제7권2호
• /
• pp.113-128
• /
• 2022

Water consumption is strongly affected by numerous factors, such as population, climatic, geographic, and socio-economic factors. Therefore, the implementation of a reliable predictive model of water consumption pattern is challenging task. This study investigates the performance of predictive models based on multi-layer perceptron (MLP), multiple linear regression (MLR), and support vector regression (SVR). To understand the significant factors affecting water consumption, the stepwise regression (SW) procedure is used in MLR to obtain suitable variables. Then, this study also implements three predictive models based on these significant variables (e.g., SWMLR, SWMLP, and SWSVR). Annual data of water consumption in Thailand during 2006 - 2015 were compiled and categorized by provinces and distributors. By comparing the predictive performance of models with all variables, the results demonstrate that the MLP models outperformed the MLR and SVR models. As compared to the models with selected variables, the predictive capability of SWMLP was superior to SWMLR and SWSVR. Therefore, the SWMLP still provided satisfactory results with the minimum number of explanatory variables which in turn reduced the computation time and other resources required while performing the predictive task. It can be concluded that the MLP exhibited the best result and can be utilized as a reliable water demand predictive model for both of all variables and selected variables cases. These findings support important implications and serve as a feasible water consumption predictive model and can be used for water resources management to produce sufficient tap water to meet the demand in each province of Thailand.

• Computers and Concrete
• /
• 제15권1호
• /
• pp.89-101
• /
• 2015

In this paper, Artificial Neural Networks (ANN) and Multiple Linear Regression (MLR) models were discussed to determine the compressive strength of clinker mortars cured for 1, 2, 7 and 28 days. In the experimental stage, 1288 mortar samples were produced from 322 different clinker specimens and compressive strength tests were performed on these samples. Chemical properties of the clinker samples were also determined. In the modeling stage, these experimental results were used to construct the models. In the models tricalcium silicate ($C_3S$), dicalcium silicate ($C_2S$), tricalcium aluminate ($C_3A$), tetracalcium alumina ferrite ($C_4AF$), blaine values, specific gravity and age of samples were used as inputs and the compressive strength of clinker samples was used as output. The approximate reasoning ability of the models compared using some statistical parameters. As a result, ANN has shown satisfying relation with experimental results and suggests an alternative approach to evaluate compressive strength estimation of clinker mortars using related inputs. Furthermore MLR model showed a poor ability to predict.

• 대기
• /
• 제31권1호
• /
• pp.29-43
• /
• 2021

In this study, two-step statistical approach including Principal Component Analysis (PCA) and Multiple Linear Regression (MLR) was employed, and main meteorological factors explaining the high-PM2.5 episodes were identified in two regions: Seoul and Busan. We first performed PCA to isolate the Principal Component (PC) that is linear combination of the meteorological variables observed at two levels: surface and 850 hPa level. The employed variables at surface are: temperature (T2m), wind speed, sea level pressure, south-north and west-east wind component and those at 850 hPa upper level variables are: south-north (v850) and west-east (u850) wind component and vertical stability. Secondly we carried out MLR analysis and verified the relationships between PM2.5 daily mean concentration and meteorological PCs. Our two-step statistical approach revealed that in Seoul, dominant factors for influencing the high PM2.5 days are mainly composed of upper wind characteristics in winter including positive u850 and negative v850, indicating that continental (or Siberian) anticyclone had a strong influence. In Busan, however, the dominant factors in explanaining in high PM2.5 concentrations were associated with high T2m and negative u850 in summer. This is suggesting that marine anticyclone had a considerable effect on Busan's high PM2.5 with high temperature which is relevant to the vigorous photochemical secondary generation. Our results of both differences and similarities between two regions derived from only statistical approaches imply the high-PM2.5 episodes in Korea show their own unique characteristics and seasonality which are mostly explainable by two layer (surface and upper) mesoscale meteorological variables.

• Structural Engineering and Mechanics
• /
• 제60권3호
• /
• pp.471-488
• /
• 2016

This study predicts the strength of rotary brace damper by analyzing a new set of probabilistic models using the usual method of multiple linear regressions (MLR) and advanced machine-learning methods of multivariate adaptive regression splines (MARS), Rotary brace damper can be easily assembled with high energy-dissipation capability. To investigate the behavior of this damper in structures, a steel frame is modeled with this device subjected to monotonic and cyclic loading. Several response parameters are considered, and the performance of damper in reducing each response is evaluated. MLR and MARS methods were used to predict the strength of this damper. Displacement was determined to be the most effective parameter of damper strength, whereas the thickness did not exhibit any effect. Adding thickness parameter as inputs to MARS and MLR models did not increase the accuracies of the models in predicting the strength of this damper. The MARS model with a root mean square error (RMSE) of 0.127 and mean absolute error (MAE) of 0.090 performed better than the MLR model with an RMSE of 0.221 and MAE of 0.181.

• Computers and Concrete
• /
• 제21권5호
• /
• pp.559-567
• /
• 2018

In this study alkali activation of Electric Arc Furnace Slag (EAFS) is studied with a comprehensive test program. Three different silicate moduli (1-1,5-2), three different sodium concentrations (4%-6%-8%) for each silicate module, two different curing conditions (45%-98% relative humidity) for each sodium concentration, two different curing temperatures ($400^{\circ}C-800^{\circ}C$) for each relative humidity condition and two different curing time (6h-12h) for each curing temperature variables are selected and their effects on compressive strength was evaluated then regression equations using multiple linear regressions methods are fitted. And then to select the best regression models confirm with using the variables, the regression models compared between itself. An Artificial Neural Network (ANN) models that use silicate moduli, sodium concentration, relative humidity, curing temperature and curing time variables, are formed. After the investigation of these ANN models' results, ANN and multiple linear regressions based models are compared with each other. After that, an explicit formula is developed with values of the ANN model. As a result of this study, the fluctuations of data set of the compressive strength were very well reflected using both of the methods, multiple linear regression with quadratic terms and ANN.

• 한국연초학회지
• /
• 제22권2호
• /
• pp.123-132
• /
• 2000

Evaluation method of soil fertility by combination of soil color characteristics and survey data from soil map as well as chemical properties was investigated on total 35 field and pot experiments. Total 35 tobacco fields including 11 fields located at Cheonweon county in Chungnam Province, 9 fields located at Goesan county in Chungbuk Province, and 15 fields located at Youngcheon county in Kyongbuk Province were selected in 1984 to cover the wide range of distribution in landscape and soil attributes. Yields of tobacco grown on the plots of both the pot and field experiment which were not applied with any fertilizer were considered as basic fertility of the soil (BFS). The BFS was estimated by 32 independent variables including 15 chemical properties, 3 color characteristics, and 14 soil survey data from soil map. Twenty-four independent variables containing 16 quantitative variables selected from 24 quantitative variables by collinearity diagnostics and 8 qualitative variables, were classified and analyzed by multiple linear regression (MLR) of REG and GLM models of SAS. Tobacco yield of field experiment showed high variations by eight times in difference between minimum and maximum yield indicating the diverse soil fertility among the experimental fields. Evaluation for the BFS by the MLR including quantitative variables was still more confidential than that by a single index and that showed more improvement of coefficient of determination ($R^2$) in pot experiment than in field experiment. Evaluation for the BFS by MLR in field experiment was still improved by adding qualitative variables as well as quantitative variables. The variability in the BFS of field experiment was explained 43.2% by quantitative variables and 67.95% by adding both the quantitative and qualitative variables compared with 21.7% by simple regression with NO$_3$-N content in soil. The regression evaluation for the best evaluation of the BFS of field experiment by MLR included NO$_3$-N content, L value, and a value of soil color as quantitative variables and available soil depth and topography as qualitative variables. Consequently, it is assumed that this approach by the MLR including both the quantitative and qualitative variables was available as an evaluation model of soil fertility for tobacco field.

• 전기학회논문지
• /
• 제65권12호
• /
• pp.2075-2083
• /
• 2016

The real time IAQ (Indoor Air Quality) management is very important for large buildings and underground facilities such as subways because poor IAQ is immediately harmful to human health. Such IAQ management requires monitoring, prediction and control in an integrated and real time manner. In this paper, we present three PM10 hourly prediction models for such realtime IAQ management as both Multiple Linear Regression (MLR) and Artificial Neural Network (ANN) models. Both MLR and ANN models show good performances between 0.76 and 0.88 with respect to R (correlation coefficient) between the measured and predicted values, but the MLR models outperform the corresponding ANN models with respect to RMSE (root mean square error).

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2006년도 춘계학술발표대회
• /
• pp.23-26
• /
• 2006

기존의 데이터 마이닝 예측 기법 중 회귀분석은 학습 단계에서 생성된 모델을 변경 없이 새로운 데이터에 적용하였다. 그러나 시계열 데이터에 모델 변경 없이 동일하게 적용하면 시간이 지남에 따라 정확도가 낮아지는 단점이 있다. 따라서 이 논문에서는 시간에 따라 변화하는 시계열데이터의 특성을 고려하여 점진적으로 회귀 모델을 갱신하는 기법을 제안한다. 이 기법은 입력되는 모든 데이터를 회귀 모델에 적용하여 점진적으로 모델을 갱신한다. 제안된 기법의 타당성은 RME(Relative Mean Error)와 RMSE(Root Mean Square Error)를 이용하여 측정하였다. 정확도 측정 실험 결과 제안 기법인 IMQR(Incremental Multiple Quadratic Regression) 기법이 MLR(Multiple Linear Regression), MQR(Multiple Quadratic Regression), SVR(Support Vector Regression) 기법에 비해 RME 가 평균 2%, RMSE 가 평균 0.02 정도 우수한 결과를 얻었다.