To transition from conventional to intelligent real estate, the real estate industry must enhance its embrace of disruptive technology. Even though the real estate auction market has grown in importance in the financi...To transition from conventional to intelligent real estate, the real estate industry must enhance its embrace of disruptive technology. Even though the real estate auction market has grown in importance in the financial, economic, and investment sectors, few artificial intelligence-based research has tried to predict the auction values of real estate in the past. According to the objectives of this research, artificial intelligence and statistical methods will be used to create forecasting models for real estate auction prices. A multiple regression model and an artificial neural network are used in conjunction with one another to build the forecasting models. For the empirical study, the study utilizes data from Ghana apartment auctions from 2016 to 2020 to anticipate auction prices and evaluate the forecasting accuracy of the various models available at the time. Compared to the conventional Multiple Regression Analysis, using artificial intelligence systems for real estate appraisal is becoming a more viable option (MRA). The Artificial Neural network model exhibits the most outstanding performance, and efficient zonal segmentation based on the auction evaluation price enhances the model’s prediction accuracy even more. There is a statistically significant difference between the two models when it comes to forecasting the values of real estate auctions.展开更多
Due to the phenomenon of abandoning wind power and photo voltage(PV)power in the“Three Northern Areas”in China,this paper presents an optimal strategy for coordinating and dispatching“source-load”in power system b...Due to the phenomenon of abandoning wind power and photo voltage(PV)power in the“Three Northern Areas”in China,this paper presents an optimal strategy for coordinating and dispatching“source-load”in power system based on multiple time scales.On the basis of the analysis of the uncertainty of wind power and PV power as well as the characteristics of load side resource dispatching,the optimal model of coordinating and dispatching“source-load”in power system based on multiple time scales is established.It can simultaneously and effectively dispatch conventional generators,wind plant,PV power station,pumped-storage power station and load side resources by optimally using three time scales:day-ahead,intra-day and real-time.According to the latest predicted information of wind power,PV power and load,the original generation schedule can be rolled and amended by using the corresponding time scale.The effectiveness of the model can be verified by a real system.The simulation results show that the proposed model can make full use of“source-load”resources to improve the ability to consume wind power and PV power of the grid-connected system.展开更多
The Madden-Julian Oscillation(MJO)is a key atmospheric component connecting global weather and climate.It func-tions as a primary source for subseasonal forecasts.Previous studies have highlighted the vital impact of ...The Madden-Julian Oscillation(MJO)is a key atmospheric component connecting global weather and climate.It func-tions as a primary source for subseasonal forecasts.Previous studies have highlighted the vital impact of oceanic processes on MJO propagation.However,few existing MJO prediction approaches adequately consider these factors.This study determines the critical region for the oceanic processes affecting MJO propagation by utilizing 22-year Climate Forecast System Reanalysis data.By intro-ducing surface and subsurface oceanic temperature within this critical region into a lagged multiple linear regression model,the MJO forecasting skill is considerably optimized.This optimization leads to a 12 h enhancement in the forecasting skill of the first principal component and efficiently decreases prediction errors for the total predictions.Further analysis suggests that,during the years in which MJO events propagate across the Maritime Continent over a more southerly path,the optimized statistical forecasting model obtains better improvements in MJO prediction.展开更多
Target tracking is very important in computer vision and related areas. It is usually difficult to accurately track fast motion target with appearance variations. Sometimes the tracking algorithms fail for heavy appea...Target tracking is very important in computer vision and related areas. It is usually difficult to accurately track fast motion target with appearance variations. Sometimes the tracking algorithms fail for heavy appearance variations. A multiple template method to track fast motion target with appearance changes is presented under the framework of appearance model with Kalman filter. Firstly, we construct a multiple template appearance model, which includes both the original template and templates affinely transformed from original one. Generally speaking, appearance variations of fast motion target can be covered by affine transformation. Therefore, the affine tr templates match the target of appearance variations better than conventional models. Secondly, we present an improved Kalman filter for approx- imate estimating the motion trail of the target and a modified similarity evaluation function for exact matching. The estimation approach can reduce time complexity of the algorithm and keep accuracy in the meantime. Thirdly, we propose an adaptive scheme for updating template set to alleviate the drift problem. The scheme considers the following differences： the weight differences in two successive frames; different types of affine transformation applied to templates. Finally, experiments demonstrate that the proposed algorithm is robust to appearance varia- tion of fast motion target and achieves real-time performance on middle/low-range computing platform.展开更多
With the increasing proportion of wind power integration, the volatility of wind power brings huge challenges to the safe and stable operation of the electric power system. At present, the indexes commonly used to eva...With the increasing proportion of wind power integration, the volatility of wind power brings huge challenges to the safe and stable operation of the electric power system. At present, the indexes commonly used to evaluate the volatility of wind power only consider its overall characteristics, such as the standard deviation of wind power, the average of power variables, etc., while ignoring the detailed volatility of wind power, that is, the features of the frequency distribution of power variables. However, how to accurately describe the detailed volatility of wind power is the key foundation to reduce its adverse influences. To address this, a quantitative method for evaluating the detailed volatility of wind power at multiple temporal-spatial scales is proposed. First, the volatility indexes which can evaluate the detailed fluctuation characteristics of wind power are presented, including the upper confidence limit, lower confidence limit and confidence interval of power variables under the certain confidence level. Then, the actual wind power data from a location in northern China is used to illustrate the application of the proposed indexes at multiple temporal(year–season–month–day) and spatial scales(wind turbine–wind turbines–wind farm–wind farms) using the calculation time windows of 10 min, 30 min, 1 h, and 4 h. Finally, the relationships between wind power forecasting accuracy and its corresponding detailed volatility are analyzed to further verify the effectiveness of the proposed indexes. The results show that the proposed volatility indexes can effectively characterize the detailed fluctuations of wind power at multiple temporal-spatial scales. It is anticipated that the results of this study will serve as an important reference for the reserve capacity planning and optimization dispatch in the electric power system which with a high proportion of renewable energy.展开更多
We introduce a novel approach to multifractal data in order to achieve transcended modeling and forecasting performances by extracting time series out of local Hurst exponent calculations at a specified scale.First,th...We introduce a novel approach to multifractal data in order to achieve transcended modeling and forecasting performances by extracting time series out of local Hurst exponent calculations at a specified scale.First,the long range and co-movement dependencies of the time series are scrutinized on time-frequency space using multiple wavelet coherence analysis.Then,the multifractal behaviors of the series are verified by multifractal de-trended fluctuation analysis and its local Hurst exponents are calculated.Additionally,root mean squares of residuals at the specified scale are procured from an intermediate step during local Hurst exponent calculations.These internally calculated series have been used to estimate the process with vector autoregressive fractionally integrated moving average(VARFIMA)model and forecasted accordingly.In our study,the daily prices of gold,silver and platinum are used for assessment.The results have shown that all metals do behave in phase movement on long term periods and possess multifractal features.Furthermore,the intermediate time series obtained during local Hurst exponent calculations still appertain the co-movement as well as multifractal characteristics of the raw data and may be successfully re-scaled,modeled and forecasted by using VARFIMA model.Conclusively,VARFIMA model have notably surpassed its univariate counterpart(ARFIMA)in all efficacious trials while re-emphasizing the importance of comovement procurement in modeling.Our study’s novelty lies in using a multifractal de-trended fluctuation analysis,along with multiple wavelet coherence analysis,for forecasting purposes to an extent not seen before.The results will be of particular significance to finance researchers and practitioners.展开更多
This research aims to improve the forecasting precision of electric quantity. It is discovered that the total electricity consumption considerably increased during the Spring Festival by the analysis of the electric q...This research aims to improve the forecasting precision of electric quantity. It is discovered that the total electricity consumption considerably increased during the Spring Festival by the analysis of the electric quantity time series from 2002 to 2007 in Shandong province. The festival factor is ascertained to be one of the important seasonal factors affecting the electric quantity fluctuations, and the multiplication model for forecasting is improved by introducing corresponding variables and parameters. The computational results indicate that the average relative error of the new model decreases from 4.31% to 1.93% and the maximum relative error from 14.05% to 6.52% compared with those of the model when the festival factor is not considered. It shows that introducing the festival factor into the multiplication model for electric quantity forecasting evidently improves the precision.展开更多
To assess whether a development strategy will be profitable enough,production forecasting is a crucial and difficult step in the process.The development history of other reservoirs in the same class tends to be studie...To assess whether a development strategy will be profitable enough,production forecasting is a crucial and difficult step in the process.The development history of other reservoirs in the same class tends to be studied to make predictions accurate.However,the permeability field,well patterns,and development regime must all be similar for two reservoirs to be considered in the same class.This results in very few available experiences from other reservoirs even though there is a lot of historical information on numerous reservoirs because it is difficult to find such similar reservoirs.This paper proposes a learn-to-learn method,which can better utilize a vast amount of historical data from various reservoirs.Intuitively,the proposed method first learns how to learn samples before directly learning rules in samples.Technically,by utilizing gradients from networks with independent parameters and copied structure in each class of reservoirs,the proposed network obtains the optimal shared initial parameters which are regarded as transferable information across different classes.Based on that,the network is able to predict future production indices for the target reservoir by only training with very limited samples collected from reservoirs in the same class.Two cases further demonstrate its superiority in accuracy to other widely-used network methods.展开更多
This paper set out to analyze and forecast the Hong Kong Interbank Interest Rate(HIBOR)for a period 2006 to 2018.The main objective of this study is to propose an appropriate time series forecasting model for HIBOR.HI...This paper set out to analyze and forecast the Hong Kong Interbank Interest Rate(HIBOR)for a period 2006 to 2018.The main objective of this study is to propose an appropriate time series forecasting model for HIBOR.HIBOR conceptually captures the interaction between demand and supply of Hong Kong dollar in the interbank market.The volatility of HIBOR reflects market sentiment,changes in underlying macroeconomic environment,random events and even political climate.Thus,the time series data of HIBOR appears to have multiple seasonality during the aforesaid period.The TBATS model,the state space modeling framework developed by De Livera,Hyndman and Snyder(2010)is adopted for this study to improve the accuracy and efficiency of the time series modeling and forecasting of HIBOR.The TBATS model incorporates Box-Cox transformations,Fourier representations with time varying coefficients,and ARMA error correction.Likelihood evaluation and analytical expressions for point forecasts and interval predictions under the assumption of Gaussian errors are derived,leading to a simple,comprehensive approach to forecasting complex seasonal time series.In addition,the trigonometric formulation is used as a means of decomposing complex seasonal time series,which helps to identify and extract seasonal components which are otherwise not apparent in the time series plot itself.The performance of the TBATS model as evaluated by measures of forecast error are presented.展开更多
Load forecasting is a critical issue for operational planning as well as grid expansion to ensure an uninterruptable electric power system. Being a small but densely populated country in South Asia, Bangladesh has man...Load forecasting is a critical issue for operational planning as well as grid expansion to ensure an uninterruptable electric power system. Being a small but densely populated country in South Asia, Bangladesh has many isolated places which are not connected to national grid yet. If concern authority opts to expand grid to those areas, they need reliable demand data for designing and dimensioning of different power system entities, e.g., capacity, overhead line capacity, tie line capacity, spinning reserve, load-shedding scheduling, etc., for reliable operation and to prevent possible obligatory redesigning. This paper represents an analysis to forecast the electricity demand of an isolated island in Bangladesh where past history of electrical load demand is not available. The analysis is based on the identification of factors, e.g., population, literacy rate, per capita income, occupation, communication, etc., on which electrical load growth of an area depends. Data has been collected from the targeted isolated area and form a grid connected area which is similar to target area from social and geographical perspective. Weights of those factors on load have been calculated by matrix inversion. Demand of the new area is forecasted using these weights factors by matrix multiplication.展开更多
In this paper, the model output machine learning (MOML) method is proposed for simulating weather consultation, which can improve the forecast results of numerical weather prediction (NWP). During weather consultation...In this paper, the model output machine learning (MOML) method is proposed for simulating weather consultation, which can improve the forecast results of numerical weather prediction (NWP). During weather consultation, the forecasters obtain the final results by combining the observations with the NWP results and giving opinions based on their experience. It is obvious that using a suitable post-processing algorithm for simulating weather consultation is an interesting and important topic. MOML is a post-processing method based on machine learning, which matches NWP forecasts against observations through a regression function. By adopting different feature engineering of datasets and training periods, the observational and model data can be processed into the corresponding training set and test set. The MOML regression function uses an existing machine learning algorithm with the processed dataset to revise the output of NWP models combined with the observations, so as to improve the results of weather forecasts. To test the new approach for grid temperature forecasts, the 2-m surface air temperature in the Beijing area from the ECMWF model is used. MOML with different feature engineering is compared against the ECMWF model and modified model output statistics (MOS) method. MOML shows a better numerical performance than the ECMWF model and MOS, especially for winter. The results of MOML with a linear algorithm, running training period, and dataset using spatial interpolation ideas, are better than others when the forecast time is within a few days. The results of MOML with the Random Forest algorithm, year-round training period, and dataset containing surrounding gridpoint information, are better when the forecast time is longer.展开更多
Group behavior forecasting is an emergent re- search and application field in social computing. Most of the existing group behavior forecasting methods have heavily re- lied on structured data which is usually hard to...Group behavior forecasting is an emergent re- search and application field in social computing. Most of the existing group behavior forecasting methods have heavily re- lied on structured data which is usually hard to obtain. To ease the heavy reliance on structured data, in this paper, we pro- pose a computational approach based on the recognition of multiple plans/intentions underlying group behavior. We fur- ther conduct human experiment to empirically evaluate the effectiveness of our proposed approach.展开更多
文摘To transition from conventional to intelligent real estate, the real estate industry must enhance its embrace of disruptive technology. Even though the real estate auction market has grown in importance in the financial, economic, and investment sectors, few artificial intelligence-based research has tried to predict the auction values of real estate in the past. According to the objectives of this research, artificial intelligence and statistical methods will be used to create forecasting models for real estate auction prices. A multiple regression model and an artificial neural network are used in conjunction with one another to build the forecasting models. For the empirical study, the study utilizes data from Ghana apartment auctions from 2016 to 2020 to anticipate auction prices and evaluate the forecasting accuracy of the various models available at the time. Compared to the conventional Multiple Regression Analysis, using artificial intelligence systems for real estate appraisal is becoming a more viable option (MRA). The Artificial Neural network model exhibits the most outstanding performance, and efficient zonal segmentation based on the auction evaluation price enhances the model’s prediction accuracy even more. There is a statistically significant difference between the two models when it comes to forecasting the values of real estate auctions.
基金Major Projects of Gansu Province(No.17ZD2GA010)Power Company Technology Projects of State Grid Corporation in Gansu Province(No.52272716000K)
文摘Due to the phenomenon of abandoning wind power and photo voltage(PV)power in the“Three Northern Areas”in China,this paper presents an optimal strategy for coordinating and dispatching“source-load”in power system based on multiple time scales.On the basis of the analysis of the uncertainty of wind power and PV power as well as the characteristics of load side resource dispatching,the optimal model of coordinating and dispatching“source-load”in power system based on multiple time scales is established.It can simultaneously and effectively dispatch conventional generators,wind plant,PV power station,pumped-storage power station and load side resources by optimally using three time scales:day-ahead,intra-day and real-time.According to the latest predicted information of wind power,PV power and load,the original generation schedule can be rolled and amended by using the corresponding time scale.The effectiveness of the model can be verified by a real system.The simulation results show that the proposed model can make full use of“source-load”resources to improve the ability to consume wind power and PV power of the grid-connected system.
基金supported by the National Key Program for Developing Basic Science(Nos.2022YFF0801702 and 2022YFE0106600)the National Natural Science Foundation of China(Nos.42175060 and 42175021)the Jiangsu Province Science Foundation(No.BK20250200302).
文摘The Madden-Julian Oscillation(MJO)is a key atmospheric component connecting global weather and climate.It func-tions as a primary source for subseasonal forecasts.Previous studies have highlighted the vital impact of oceanic processes on MJO propagation.However,few existing MJO prediction approaches adequately consider these factors.This study determines the critical region for the oceanic processes affecting MJO propagation by utilizing 22-year Climate Forecast System Reanalysis data.By intro-ducing surface and subsurface oceanic temperature within this critical region into a lagged multiple linear regression model,the MJO forecasting skill is considerably optimized.This optimization leads to a 12 h enhancement in the forecasting skill of the first principal component and efficiently decreases prediction errors for the total predictions.Further analysis suggests that,during the years in which MJO events propagate across the Maritime Continent over a more southerly path,the optimized statistical forecasting model obtains better improvements in MJO prediction.
基金Supported by the National Science Foundation of China(61472289)Hubei Province Science Foundation(2015CFB254)
文摘Target tracking is very important in computer vision and related areas. It is usually difficult to accurately track fast motion target with appearance variations. Sometimes the tracking algorithms fail for heavy appearance variations. A multiple template method to track fast motion target with appearance changes is presented under the framework of appearance model with Kalman filter. Firstly, we construct a multiple template appearance model, which includes both the original template and templates affinely transformed from original one. Generally speaking, appearance variations of fast motion target can be covered by affine transformation. Therefore, the affine tr templates match the target of appearance variations better than conventional models. Secondly, we present an improved Kalman filter for approx- imate estimating the motion trail of the target and a modified similarity evaluation function for exact matching. The estimation approach can reduce time complexity of the algorithm and keep accuracy in the meantime. Thirdly, we propose an adaptive scheme for updating template set to alleviate the drift problem. The scheme considers the following differences： the weight differences in two successive frames; different types of affine transformation applied to templates. Finally, experiments demonstrate that the proposed algorithm is robust to appearance varia- tion of fast motion target and achieves real-time performance on middle/low-range computing platform.
基金supported in part by the National Key R&D Program of China (No.2017YFE0109000)the project of China Datang Corporation Ltd
文摘With the increasing proportion of wind power integration, the volatility of wind power brings huge challenges to the safe and stable operation of the electric power system. At present, the indexes commonly used to evaluate the volatility of wind power only consider its overall characteristics, such as the standard deviation of wind power, the average of power variables, etc., while ignoring the detailed volatility of wind power, that is, the features of the frequency distribution of power variables. However, how to accurately describe the detailed volatility of wind power is the key foundation to reduce its adverse influences. To address this, a quantitative method for evaluating the detailed volatility of wind power at multiple temporal-spatial scales is proposed. First, the volatility indexes which can evaluate the detailed fluctuation characteristics of wind power are presented, including the upper confidence limit, lower confidence limit and confidence interval of power variables under the certain confidence level. Then, the actual wind power data from a location in northern China is used to illustrate the application of the proposed indexes at multiple temporal(year–season–month–day) and spatial scales(wind turbine–wind turbines–wind farm–wind farms) using the calculation time windows of 10 min, 30 min, 1 h, and 4 h. Finally, the relationships between wind power forecasting accuracy and its corresponding detailed volatility are analyzed to further verify the effectiveness of the proposed indexes. The results show that the proposed volatility indexes can effectively characterize the detailed fluctuations of wind power at multiple temporal-spatial scales. It is anticipated that the results of this study will serve as an important reference for the reserve capacity planning and optimization dispatch in the electric power system which with a high proportion of renewable energy.
文摘We introduce a novel approach to multifractal data in order to achieve transcended modeling and forecasting performances by extracting time series out of local Hurst exponent calculations at a specified scale.First,the long range and co-movement dependencies of the time series are scrutinized on time-frequency space using multiple wavelet coherence analysis.Then,the multifractal behaviors of the series are verified by multifractal de-trended fluctuation analysis and its local Hurst exponents are calculated.Additionally,root mean squares of residuals at the specified scale are procured from an intermediate step during local Hurst exponent calculations.These internally calculated series have been used to estimate the process with vector autoregressive fractionally integrated moving average(VARFIMA)model and forecasted accordingly.In our study,the daily prices of gold,silver and platinum are used for assessment.The results have shown that all metals do behave in phase movement on long term periods and possess multifractal features.Furthermore,the intermediate time series obtained during local Hurst exponent calculations still appertain the co-movement as well as multifractal characteristics of the raw data and may be successfully re-scaled,modeled and forecasted by using VARFIMA model.Conclusively,VARFIMA model have notably surpassed its univariate counterpart(ARFIMA)in all efficacious trials while re-emphasizing the importance of comovement procurement in modeling.Our study’s novelty lies in using a multifractal de-trended fluctuation analysis,along with multiple wavelet coherence analysis,for forecasting purposes to an extent not seen before.The results will be of particular significance to finance researchers and practitioners.
基金The Forecasting Research Base of Chinese Academy of Sciences in Xi an Jiaotong University,the National Natural Science Foundation of China (No.70773091)
文摘This research aims to improve the forecasting precision of electric quantity. It is discovered that the total electricity consumption considerably increased during the Spring Festival by the analysis of the electric quantity time series from 2002 to 2007 in Shandong province. The festival factor is ascertained to be one of the important seasonal factors affecting the electric quantity fluctuations, and the multiplication model for forecasting is improved by introducing corresponding variables and parameters. The computational results indicate that the average relative error of the new model decreases from 4.31% to 1.93% and the maximum relative error from 14.05% to 6.52% compared with those of the model when the festival factor is not considered. It shows that introducing the festival factor into the multiplication model for electric quantity forecasting evidently improves the precision.
基金This work is supported by the National Natural Science Foundation of China under Grant 52274057,52074340 and 51874335the Major Scientific and Technological Projects of CNPC under Grant ZD2019-183-008+2 种基金the Major Scientific and Technological Projects of CNOOC under Grant CCL2022RCPS0397RSNthe Science and Technology Support Plan for Youth Innovation of University in Shandong Province under Grant 2019KJH002111 Project under Grant B08028.
文摘To assess whether a development strategy will be profitable enough,production forecasting is a crucial and difficult step in the process.The development history of other reservoirs in the same class tends to be studied to make predictions accurate.However,the permeability field,well patterns,and development regime must all be similar for two reservoirs to be considered in the same class.This results in very few available experiences from other reservoirs even though there is a lot of historical information on numerous reservoirs because it is difficult to find such similar reservoirs.This paper proposes a learn-to-learn method,which can better utilize a vast amount of historical data from various reservoirs.Intuitively,the proposed method first learns how to learn samples before directly learning rules in samples.Technically,by utilizing gradients from networks with independent parameters and copied structure in each class of reservoirs,the proposed network obtains the optimal shared initial parameters which are regarded as transferable information across different classes.Based on that,the network is able to predict future production indices for the target reservoir by only training with very limited samples collected from reservoirs in the same class.Two cases further demonstrate its superiority in accuracy to other widely-used network methods.
基金a grant from the College of Professional and Continuing Education,an affiliate of The Hong Kong Polytechnic University。
文摘This paper set out to analyze and forecast the Hong Kong Interbank Interest Rate(HIBOR)for a period 2006 to 2018.The main objective of this study is to propose an appropriate time series forecasting model for HIBOR.HIBOR conceptually captures the interaction between demand and supply of Hong Kong dollar in the interbank market.The volatility of HIBOR reflects market sentiment,changes in underlying macroeconomic environment,random events and even political climate.Thus,the time series data of HIBOR appears to have multiple seasonality during the aforesaid period.The TBATS model,the state space modeling framework developed by De Livera,Hyndman and Snyder(2010)is adopted for this study to improve the accuracy and efficiency of the time series modeling and forecasting of HIBOR.The TBATS model incorporates Box-Cox transformations,Fourier representations with time varying coefficients,and ARMA error correction.Likelihood evaluation and analytical expressions for point forecasts and interval predictions under the assumption of Gaussian errors are derived,leading to a simple,comprehensive approach to forecasting complex seasonal time series.In addition,the trigonometric formulation is used as a means of decomposing complex seasonal time series,which helps to identify and extract seasonal components which are otherwise not apparent in the time series plot itself.The performance of the TBATS model as evaluated by measures of forecast error are presented.
文摘Load forecasting is a critical issue for operational planning as well as grid expansion to ensure an uninterruptable electric power system. Being a small but densely populated country in South Asia, Bangladesh has many isolated places which are not connected to national grid yet. If concern authority opts to expand grid to those areas, they need reliable demand data for designing and dimensioning of different power system entities, e.g., capacity, overhead line capacity, tie line capacity, spinning reserve, load-shedding scheduling, etc., for reliable operation and to prevent possible obligatory redesigning. This paper represents an analysis to forecast the electricity demand of an isolated island in Bangladesh where past history of electrical load demand is not available. The analysis is based on the identification of factors, e.g., population, literacy rate, per capita income, occupation, communication, etc., on which electrical load growth of an area depends. Data has been collected from the targeted isolated area and form a grid connected area which is similar to target area from social and geographical perspective. Weights of those factors on load have been calculated by matrix inversion. Demand of the new area is forecasted using these weights factors by matrix multiplication.
基金supported by the National Key Research and Development Program of China (Grant Nos. 2018YFF0300104 and 2017YFC0209804)the National Natural Science Foundation of China (Grant No. 11421101)Beijing Academy of Artifical Intelligence (BAAI)
文摘In this paper, the model output machine learning (MOML) method is proposed for simulating weather consultation, which can improve the forecast results of numerical weather prediction (NWP). During weather consultation, the forecasters obtain the final results by combining the observations with the NWP results and giving opinions based on their experience. It is obvious that using a suitable post-processing algorithm for simulating weather consultation is an interesting and important topic. MOML is a post-processing method based on machine learning, which matches NWP forecasts against observations through a regression function. By adopting different feature engineering of datasets and training periods, the observational and model data can be processed into the corresponding training set and test set. The MOML regression function uses an existing machine learning algorithm with the processed dataset to revise the output of NWP models combined with the observations, so as to improve the results of weather forecasts. To test the new approach for grid temperature forecasts, the 2-m surface air temperature in the Beijing area from the ECMWF model is used. MOML with different feature engineering is compared against the ECMWF model and modified model output statistics (MOS) method. MOML shows a better numerical performance than the ECMWF model and MOS, especially for winter. The results of MOML with a linear algorithm, running training period, and dataset using spatial interpolation ideas, are better than others when the forecast time is within a few days. The results of MOML with the Random Forest algorithm, year-round training period, and dataset containing surrounding gridpoint information, are better when the forecast time is longer.
文摘Group behavior forecasting is an emergent re- search and application field in social computing. Most of the existing group behavior forecasting methods have heavily re- lied on structured data which is usually hard to obtain. To ease the heavy reliance on structured data, in this paper, we pro- pose a computational approach based on the recognition of multiple plans/intentions underlying group behavior. We fur- ther conduct human experiment to empirically evaluate the effectiveness of our proposed approach.
