• Title/Summary/Keyword: RMSE

Search Result 1,793, Processing Time 0.033 seconds

Forecasting Fish Import Using Deep Learning: A Comprehensive Analysis of Two Different Fish Varieties in South Korea

  • Abhishek Chaudhary;Sunoh Choi
    • Smart Media Journal
    • /
    • v.12 no.11
    • /
    • pp.134-144
    • /
    • 2023
  • Nowadays, Deep Learning (DL) technology is being used in several government departments. South Korea imports a lot of seafood. If the demand for fishery products is not accurately predicted, then there will be a shortage of fishery products and the price of the fishery product may rise sharply. So, South Korea's Ministry of Ocean and Fisheries is attempting to accurately predict seafood imports using deep learning. This paper introduces the solution for the fish import prediction in South Korea using the Long Short-Term Memory (LSTM) method. It was found that there was a huge gap between the sum of consumption and export against the sum of production especially in the case of two species that are Hairtail and Pollock. An import prediction is suggested in this research to fill the gap with some advanced Deep Learning methods. This research focuses on import prediction using Machine Learning (ML) and Deep Learning methods to predict the import amount more precisely. For the prediction, two Deep Learning methods were chosen which are Artificial Neural Network (ANN) and Long Short-Term Memory (LSTM). Moreover, the Machine Learning method was also selected for the comparison between the DL and ML. Root Mean Square Error (RMSE) was selected for the error measurement which shows the difference between the predicted and actual values. The results obtained were compared with the average RMSE scores and in terms of percentage. It was found that the LSTM has the lowest RMSE score which showed the prediction with higher accuracy. Meanwhile, ML's RMSE score was higher which shows lower accuracy in prediction. Moreover, Google Trend Search data was used as a new feature to find its impact on prediction outcomes. It was found that it had a positive impact on results as the RMSE values were lowered, increasing the accuracy of the prediction.

Improving the Usage of the Korea Meteorological Administration's Digital Forecasts in Agriculture: V. Field Validation of the Sky-condition based Lapse Rate Estimation Scheme (기상청 동네예보의 영농활용도 증진을 위한 방안: V. 하늘상태 기반 기온감률 추정기법의 실용성 평가)

  • Kim, Soo-ock;Yun, Jin I.
    • Korean Journal of Agricultural and Forest Meteorology
    • /
    • v.18 no.3
    • /
    • pp.135-142
    • /
    • 2016
  • The aim of this study was to confirm the improvement of efficiency for temperature estimation at 0600 and 1500 LST by using a simple method for estimating temperature lapse rate modulated by the amount of clouds in comparison with the case adopting the existing single temperature lapse rate ($-6.5^{\circ}C/km$ or $-9^{\circ}C/km$). A catchment of the 'Hadong Watermark2,' which includes Hadong, Gurye, and Gwangyang was selected as the area for evaluating the practicality of the temperature lapse rate estimation method. The weather data of 0600 and 1500 LST at 12 weather observation sites within the catchment were collected during the entire year of 2015. Also, the 'sky condition' of digital forecast products of KMA in 2015 ($5{\times}5km$ lattice resolution) were overlapped with the catchment of the 'Hadong Watermark2,' to calculate the spatial average value within the catchment, which were used to simulate the 0600 and 1500 LST temperature lapse rate of the catchment. The estimation errors of the temperatures at 0600 LST were ME $-0.39^{\circ}C$ and RMSE $1.45^{\circ}C$ in 2015, when applying the existing temperature lapse rate. Using the estimated temperature lapse rate, they were improved to ME $-0.19^{\circ}C$ and RMSE $1.32^{\circ}C$. At 1500 LST, the effect of the improvements found from the comparison between the existing temperature lapse rate and the estimated temperature lapse rate were minute, because the estimated lapse rate of clear days is not very different from the existing lapse rate. However, the estimation errors of the temperatures at 1500 LST during cloudy days were improved from ME $-0.69^{\circ}C$, RMSE $1.54^{\circ}C$ to ME $-0.51^{\circ}C$, RMSE $1.19^{\circ}C$.

Applying Nonlinear Mixed-effects Models to Taper Equations: A Case Study of Pinus densiflora in Gangwon Province, Republic of Korea (비선형 혼합효과 모형의 수간곡선 적용: 강원지방 소나무를 대상으로)

  • Shin, Joong-Hoon;Han, Hee;Ko, Chi-Ung;Kang, Jin-Taek;Kim, Young-Hwan
    • Journal of Korean Society of Forest Science
    • /
    • v.111 no.1
    • /
    • pp.136-149
    • /
    • 2022
  • In this study, the performance of a nonlinear mixed-effects (NLME) model used to estimate the stem taper of Pinus densiflora in Gangwon Province was compared with that of a nonlinear fixed-effects (NLFE) model using several performance measures. For the diameters of whole tree stems, the NLME model improved on the performance of the NLFE model by 26.4%, 42.9%, 43.1%, and 0.9% in terms of BIAS, MAB, RMSE, and FI, respectively. For the cross-section areas of whole tree stems, the NLME model improved on the performance of the NLFE model by 67.7%, 44.7%, 45.8%, and 1.0% in terms of BIAS, MAB, RMSE, and FI, respectively. Based on the analysis of 12 relative height classes of tree stems, stem taper estimation performance was also reasonably improved by the NLME model, which showed better MAB, RMSE, and FI at every relative height class compared with those of the NLFE model. In some classes, the NLFE model had better BIAS than the NLME model (stem diameter: 0.05, 0.2, 0.3, and 0.8; stem cross-section area: 0.05, 0.3, 0.5, 0.6, and 1.0). However, the NLME model enhanced the performance of stem diameter and cross-section area estimations at the lowest stem part (0.2 m from the ground). Improvements for stem diameter in terms of BIAS, MAB, RMSE, and FI were 84.2%, 69.8%, 68.7%, and 3.1%, respectively. For stem cross-section areas, the improvements in BIAS, MAB, RMSE, and FI were 98.5%, 70.1%, 68.7%, and 3.1%, respectively. The cross-section area at 0.2 m from the ground occupied 22.7% of total cross-section area. Improvements in estimation of cross-section area at the lowest stem part indicate that stem volume estimation performance could also be enhanced. Although NLME models are more difficult to fit than NLFE models, the use of NLME models as a standard method for the estimating the parameters of stem taper equations should be considered.

Shape From Focus Algorithm with Optimization of Focus Measure for Cell Image (초점 연산자의 최적화를 통한 세포영상의 삼차원 형상 복원 알고리즘)

  • Lee, Ik-Hyun;Choi, Tae-Sun
    • The Journal of Korea Institute of Information, Electronics, and Communication Technology
    • /
    • v.3 no.3
    • /
    • pp.8-13
    • /
    • 2010
  • Shape form focus (SFF) is a technique that reconstructs 3D shape of an object using image focus. Although many SFF methods have been proposed, there are still notable inaccuracy effects due to noise and non-optimization of image characteristics. In this paper, we propose a noise filter technique for noise reduction and genetic algorithm (GA) for focus measure optimization. The proposed method is analyzed with a statistical criteria such as Root Mean Square Error (RMSE) and correlation.

  • PDF

Correction of UAV's Position/Altitude through Aerial Triangulation (Aerial Triangulation을 이용한 UAV의 위치/자세 보정)

  • Choi, Kyoung-Ah;Lee, Im-Pyeong
    • Proceedings of the KSRS Conference
    • /
    • 2009.03a
    • /
    • pp.61-65
    • /
    • 2009
  • 매년 재난/재해의 발생 빈도와 피해 규모가 증가하고 있다. 그 피해를 최소화하기 위해 주기적인 모니터링을 수행하여 위기 상황을 사전에 대비하고 긴급 대응 체계를 구축하여 상황 발생 시 피해 상황을 신속하게 파악할 수시스템에 있어야 한다. 모니터링의 용이성과 신속성을 확보하기 위해 UAV에 기반한 긴급 매핑 대한 관심이 증가하고 있다. 그러나 이러한 시스템으로부터 획득된 센서 데이터가 Georeferencing되었을 때 이로부터 다양한 공간 정보를 도출할 수 있다 본 논문에서는 UAV 기반의 매핑 시스템으로부터 획득된 센서 데이터를 시뮬레이션 해보고 시뮬레이션 데이터에 대하여 Aerial Triangulation을 수행하여 영상을 Georeferncing하고 위치/자세 정보를 보정하고자 한다. 실험은 (1) 시뮬레이션 데이터 생성, (2) 초기값 생성, (3) AT 수행을 통한 위치/자세 조정의 3단계로 구성된다. 800m 길이의 1개 스트립, 500m 길이의 2개 스트립으로 나눠 비행경로를 정하고 200m, 400m, 600m의 비행고도에 대하여 다양한 실험을 수행하였다. 실험 결과 위치/자세의 초기값 RMSE에서 90% 이상 개선된 RMSE를 얻을 수 있었으며, 비행고도가 높아질수록 RMSE의 향상도는 반비례하였다. 향후에는 Sequential 알고리즘을 적용하여 연산 속도를 향상시킬 수 있고 궁극적으로 실시간 영상 Georeferencing을 가능하게 할 것으로 기대된다.

  • PDF

An alternative method for estimating lognormal means

  • Kwon, Yeil
    • Communications for Statistical Applications and Methods
    • /
    • v.28 no.4
    • /
    • pp.351-368
    • /
    • 2021
  • For a probabilistic model with positively skewed data, a lognormal distribution is one of the key distributions that play a critical role. Several lognormal models can be found in various areas, such as medical science, engineering, and finance. In this paper, we propose a new estimator for a lognormal mean and depict the performance of the proposed estimator in terms of the relative mean squared error (RMSE) compared with Shen's estimator (Shen et al., 2006), which is considered the best estimator among the existing methods. The proposed estimator includes a tuning parameter. By finding the optimal value of the tuning parameter, we can improve the average performance of the proposed estimator over the typical range of σ2. The bias reduction of the proposed estimator tends to exceed the increased variance, and it results in a smaller RMSE than Shen's estimator. A numerical study reveals that the proposed estimator has performance comparable with Shen's estimator when σ2 is small and exhibits a meaningful decrease in the RMSE under moderate and large σ2 values.

Estimation of Surface Solar Radiation using Ground-based Remote Sensing Data on the Seoul Metropolitan Area (수도권지역의 지상기반 원격탐사자료를 이용한 지표면 태양에너지 산출)

  • Jee, Joon-Bum;Min, Jae-Sik;Lee, Hankyung;Chae, Jung-Hoon;Kim, Sangil
    • Journal of the Korean earth science society
    • /
    • v.39 no.3
    • /
    • pp.228-240
    • /
    • 2018
  • Solar energy is calculated using meteorological (14 station), ceilometer (2 station) and microwave radiometer (MWR, 7 station)) data observed from the Weather Information Service Engine (WISE) on the Seoul metropolitan area. The cloud optical thickness and the cloud fraction are calculated using the back-scattering coefficient (BSC) of the ceilometer and liquid water path of the MWR. The solar energy on the surface is calculated using solar radiation model with cloud fraction from the ceilometer and the MWR. The estimated solar energy is underestimated compared to observations both at Jungnang and Gwanghwamun stations. In linear regression analysis, the slope is less than 0.8 and the bias is negative which is less than $-20W/m^2$. The estimated solar energy using MWR is more improved (i.e., deterministic coefficient (average $R^2=0.8$) and Root Mean Square Error (average $RMSE=110W/m^2$)) than when using ceilometer. The monthly cloud fraction and solar energy calculated by ceilometer is greater than 0.09 and lower than $50W/m^2$ compared to MWR. While there is a difference depending on the locations, RMSE of estimated solar radiation is large over $50W/m^2$ in July and September compared to other months. As a result, the estimation of a daily accumulated solar radiation shows the highest correlation at Gwanghwamun ($R^2=0.80$, RMSE=2.87 MJ/day) station and the lowest correlation at Gooro ($R^2=0.63$, RMSE=4.77 MJ/day) station.

Improvement of KOMPSAT Imagery Locational Accuracy Using Value-Added Processing System (부가처리시스템을 이용한 다목적실용위성 영상자료 위치정확도 개선)

  • LEE, Kwang-Jae;YUN, Hee-Cheon;KIM, Youn-Soo
    • Journal of the Korean Association of Geographic Information Studies
    • /
    • v.18 no.4
    • /
    • pp.68-80
    • /
    • 2015
  • To increase the utilization of the KOrea Multi-Purpose SATellite(KOMPSAT) series imagery being developed pursuant to the national space development program, high quality images with enhanced locational accuracy should be created through standardized post-processing processes. In the present study, using the Value-Added Processing System(VAPS) constructed for the post-processing of KOMPSAT imagery, location correction experiments were conducted using KOMPSAT-2 and -3 imagery from domestic and overseas regions. First, 50 pieces from each of KOMPSAT-2 imagery were selected from South Korean and North Korean regions, and modeling was conducted using GCP Chips. According to the results, the Root Mean Square Errors(RMSE) for South Korea and North Korea were 1.59 pixels and 2.04 pixels, respectively, and the locational accuracy of ortho mosaic imagery using check points were 1.33m(RMSE) and 1.90m(RMSE), respectively. Meanwhile, in the case of overseas regions for which GCP could not be easily obtained, the improvement of locational accuracy could be identified through image corrections using Open Street Map(OSM). The VAPS and reference materials used in the present study are expected to be very useful in constructing a precise image DB for entire global regions.

Assessment of Possibility of Adopting the Error Tolerance of Geometric Correction on Producing 1/5,000 Digital Topographic Map for Unaccessible Area Using the PLEIADES Images and TerraSAR Control Point (PLEIADES 영상과 TerraSAR 기준점을 활용한 비접근지역의 1/5,000 수치지형도 제작을 위한 기하보정의 허용오차 만족 가능성 평가)

  • Jin Kyu, Shin;Young Jin, Lee;Gyung Jong, Kim;Jun Hyuk, Lee
    • Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
    • /
    • v.33 no.2
    • /
    • pp.83-94
    • /
    • 2015
  • Recently, the necessity of spatial data in unaccessible area was challenged to set up various plans and policies for preparing the unification and the cooperative projects between South-North Korea. Therefore, this paper planned to evaluate the possibility of adopting the error tolerance in Geometric correction for 1/5,000 digital topographic mapping, using the PLEIADES images and the TerraSAR GCPs (Ground Control Points). The geometric correction was performed by changing the number and placement of GCPs by GPS (Global Positioning System) surveying, as the optimal placement of 5 GCPs were selected considering the geometric stability and steady rate. The positional accuracy evaluated by the TerraSAR GCPs, which were selected by optimal placement of GCPs. The RMSE in control points were X=±0.64m, Y=±0.46m, Z=±0.28m. While the result of geometric correction for PLEIADES images confirmed that the RMSE in control points were X=±0.34m, Y=±0.27m, Z=±0.11m, the RMSE in check points were X=±0.50m, Y=±0.30m, Z=±0.66m. Through this study, we believe if spatial data can integrate with the PLEIADES images and the optimal TerraSAR GCPs, it will be able to obtain the high-precision spatial data for adopting the regulation of 1/5,000 digital topographic map, which adjusts the computation as well as the error bound.

Integrating UAV Remote Sensing with GIS for Predicting Rice Grain Protein

  • Sarkar, Tapash Kumar;Ryu, Chan-Seok;Kang, Ye-Seong;Kim, Seong-Heon;Jeon, Sae-Rom;Jang, Si-Hyeong;Park, Jun-Woo;Kim, Suk-Gu;Kim, Hyun-Jin
    • Journal of Biosystems Engineering
    • /
    • v.43 no.2
    • /
    • pp.148-159
    • /
    • 2018
  • Purpose: Unmanned air vehicle (UAV) remote sensing was applied to test various vegetation indices and make prediction models of protein content of rice for monitoring grain quality and proper management practice. Methods: Image acquisition was carried out by using NIR (Green, Red, NIR), RGB and RE (Blue, Green, Red-edge) camera mounted on UAV. Sampling was done synchronously at the geo-referenced points and GPS locations were recorded. Paddy samples were air-dried to 15% moisture content, and then dehulled and milled to 92% milling yield and measured the protein content by near-infrared spectroscopy. Results: Artificial neural network showed the better performance with $R^2$ (coefficient of determination) of 0.740, NSE (Nash-Sutcliffe model efficiency coefficient) of 0.733 and RMSE (root mean square error) of 0.187% considering all 54 samples than the models developed by PR (polynomial regression), SLR (simple linear regression), and PLSR (partial least square regression). PLSR calibration models showed almost similar result with PR as 0.663 ($R^2$) and 0.169% (RMSE) for cloud-free samples and 0.491 ($R^2$) and 0.217% (RMSE) for cloud-shadowed samples. However, the validation models performed poorly. This study revealed that there is a highly significant correlation between NDVI (normalized difference vegetation index) and protein content in rice. For the cloud-free samples, the SLR models showed $R^2=0.553$ and RMSE = 0.210%, and for cloud-shadowed samples showed 0.479 as $R^2$ and 0.225% as RMSE respectively. Conclusion: There is a significant correlation between spectral bands and grain protein content. Artificial neural networks have the strong advantages to fit the nonlinear problem when a sigmoid activation function is used in the hidden layer. Quantitatively, the neural network model obtained a higher precision result with a mean absolute relative error (MARE) of 2.18% and root mean square error (RMSE) of 0.187%.