Deterministic Ensemble Forecasts Using Gene-Expression Programming*

Bakhshaii, Atoossa; Stull, Roland B.

doi:10.1175/2009waf2222192.1

Cited by 39 publications

(31 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The results obtained from application of GP to data from the Chute-du-Diable weather station in North Eastern Canada outperformed benchmark results from commonly used statistical downscaling models. GP has also been used for climate prediction problems including rainfall-runoff modelling [26], groundwater level fluctuations [27], short-term temperature prediction [28] and CO 2 emission modelling [29], the combination of ensemble forecasts [30], the forecasting of El Nino [31], evapotranspiration modelling (the process by which water is lost to the atmosphere from the ground surface via evaporation and plant transpiration) [32], modelling the relationship between solar activity and earth temperature [33], stream flow forecasting (forecasting of stream flow rate in a river) [34], modelling of monthly mean maximum temperature [35], modelling of water temperature [36], and wind prediction [37]. Hence we can see that there has been fairly widespread use of GP in this domain, although no previous application to the problem of seasonal forecasting was noted.…”

Section: Genetic Programming In Time-series Modellingmentioning

confidence: 99%

Genetic Programming for the Induction of Seasonal Forecasts: A Study on Weather Derivatives

Agapitos

O’Neill

Brabazon

2012

Springer Optimization and Its Applications

View full text Add to dashboard Cite

Publication informationDoumpos, M., Zopounidis, C. Abstract. The last ten years has seen the introduction and rapid growth of a market in weather derivatives, financial instruments whose payoffs are determined by the outcome of an underlying weather metric. These instruments allow organisations to protect themselves against the commercial risks posed by weather fluctuations and also provide investment opportunities for financial traders. The size of the market for weather derivatives is substantial, with a survey suggesting that the market size exceeded $45.2 Billion in 2005/2006 with most contracts being written on temperature-based metrics. A key problem faced by buyers and sellers of weather derivatives is the determination of an appropriate pricing model (and resulting price) for the financial instrument. A critical input into the pricing model is an accurate forecast of the underlying weather metric. In this study we induce seasonal forecasting temperature models by means of a Machine Learning algorithm. Genetic Programming (GP) is applied to learn an accurate, localised, long-term forecast of a temperature profile as part of the broader process of determining appropriate pricing model for weather-derivatives. Two different approaches for GP-based time-series modelling are adopted. The first is based on a simple system identification approach whereby the temporal index of the time-series is used as the sole regressor of the evolved model. The second is based on iterated single-step prediction that resembles autoregressive and moving average models in statistical time-series modelling. The major issue of effective model generalisation is tackled though the use of an ensemble learning technique that allows a family of forecasting models to be evolved using different training sets, so that predictions are formed by averaging the diverse model outputs. Empirical results suggest that GP is able to successfully induce seasonal forecasting models, and that search-based autoregressive models compose a more stable unit of evolution in terms of generalisation performance for the three datasets considered. In addition, the use of ensemble learning of 5-model predictors enhanced the generalisation ability of the system as opposed to single-model prediction systems. On a more general note, there is an increasing recognition of the utility of evolutionary methodologies for the 2 A l e x a n d r o sA g a p i t o s ,M i c h a e lO ' N e i l l ,A n t h o n yB r a b a z o n modelling of meteorological, climatic and ecological phenomena, and this work also contributes to this literature.

show abstract

Section: Genetic Programming In Time-series Modellingmentioning

confidence: 99%

Genetic Programming for the Induction of Seasonal Forecasts: A Study on Weather Derivatives

Agapitos

O’Neill

Brabazon

2012

Springer Optimization and Its Applications

View full text Add to dashboard Cite

show abstract

“…What makes GEP different from previous genetic programming methods is that the genome is coded by reading the expression tree like a book, rather than by following the node connections. Decoding back to an algorithm is possible because the arity of each operator and each basic function are known (Ferreira, 2006;Bakhshaii and Stull, 2009). The read-like-a-book coding is the key to the efficiency of GEP, because every mutation gives a viable individual (i.e., a mathematical expression that can be evaluated).…”

Section: Electric Load Forecasting For Western Canada / 353mentioning

confidence: 99%

“…A new variant, called gene expression programming (GEP;Ferreira, 2006), is very efficient at finding solutions. GEP has been used in load forecasting (Sadat Hosseini and Gandomi, 2010;Bakhshaii and Stull, 2011), numerical weather prediction (Bakhshaii and Stull, 2009;Roebber, 2010;Stull, 2011), hydrology (Aytek et al, 2008) and many other fields in recent years. This work examines hourly electrical load forecasting using GEP and artificial neural networks (ANN).…”

Section: Introductionmentioning

confidence: 99%

Electric Load Forecasting for Western Canada: A Comparison of Two Non-Linear Methods

Bakhshaii

Stull

2012

Atmosphere-Ocean

View full text Add to dashboard Cite

Seven years of hourly temperature and electric load data for British Columbia in western Canada were used to compare two statistical methods, artificial neural networks (ANN) and gene expression programming (GEP), to produce hour-ahead load forecasts. Two linear control methods (persistence and multiple linear regression) were used for verification purposes. A two-stage (predictor-corrector) approach was used. The first stage used a single regression model that applied weather and calendar data for the previous hour to predict load for any hour of the day, and the second stage applied different corrections every hour, based on a high correlation of today's load with yesterday's load for any given hour. By excluding the day-before variables, the two-stage method reduced the total number of variables in the first-stage regression and gave better results. The first five years of data were used for training (regression finding) and validation (comparative testing of candidate functions to reduce overfitting) and the last two for verification (scoring with independent data).It was found that both non-linear methods worked better than the linear methods for the first stage. All methods worked well for the second stage; hence, persistence is recommended for the second stage because it is easiest to do. After both stages, the load error was less than 0.6% of the load magnitude for hour-ahead forecasts. When used iteratively to create forecasts up to 24 hours ahead, errors grew to about 3.5% of the load magnitude for both GEP and ANNs. We also experimented by training the statistical methods with a shorter period (one year) of past data to examine the over-fitting problem. Overall, ANNs were more useful in fitting the curves to robust data, and GEP was superior for short datasets and was less sensitive to the length of the dataset. We also found that a time-nested electric load forecasting model with different lead times kept maximum load errors to 2.1% of the load magnitude out to a horizon of 24 forecasts.RÉSUMÉ [Traduit par la rédaction] Nous avons utilisé sept années de données de température et de charge électrique pour la Colombie-Britannique dans l'ouest du Canada pour comparer deux méthodes statistiques, les réseaux neuronaux artificiels (ANN) et la programmation génétique (GEP), pour produire des prévisions de charge une heure à l'avance. Nous nous sommes servis de deux méthodes de vérification linéaires (persistance et régression linéaire multiple) à des fins de vérification. Nous avons utilisé une approche en deux étapes (prédicteur-correcteur). Dans la première étape, un modèle de régression simple appliquait des données météorologiques et de calendrier pour l'heure précédente pour prévoir la charge à n'importe quelle heure de la journée et la deuxième étape appliquait à chaque heure différentes corrections basées sur une forte corrélation de la charge d'aujourd'hui avec la charge d'hier pour une heure quelconque. En excluant les variables du jour précédent, la méthode à deux étapes a réduit le nombr...

show abstract

“…An important goal of ensemble prediction is to provide estimations of the reliability of the forecast being made (Kalnay, 2003;Grimit and Mass, 2005). The ensemble of forecasts from single or multiple numerical weather prediction models provides detail of the forecast, regarding the confidence, possible errors and probability outcomes (Bakhshaii and Stull, 2009).…”

Section: Introductionmentioning

confidence: 99%

A short-range weather prediction system for South Africa based on a multi-model approach

Landman

Engelbrecht

et al. 2012

WSA

View full text Add to dashboard Cite

The accurate prediction of rainfall events, in terms of their timing, location and rainfall depth, is important to a wide range of social and economic applications. At many operational weather prediction centres, as is also the case at the South African Weather Service, forecasters use deterministic model outputs as guidance to produce subjective probabilistic rainfall forecasts. The aim of this research was to determine the skill of a new objective multi-model, multi-institute probabilistic ensemble forecast system for South Africa. Such forecasts are obtained by combining the rainfall forecasts of 2 operational high-resolution regional atmospheric models in South Africa. The first model is the Unified Model (UM), which is operational at the South African Weather Service. The UM contributes 3 ensemble members, each with a different physics scheme, data assimilation techniques and horizontal resolution. The second model is the Conformal-Cubic Atmospheric Model (CCAM) which is operational at the Council for Scientific and Industrial Research, which in turn contributed 2 members to the ensemble system based on different horizontal resolutions. A single-model ensemble forecast, with each of the ensemble members having equal weights, was constructed for the UM and CCAM models, respectively. These UM and CCAM single-model ensemble predictions are then combined into a multi-model ensemble prediction, using simple un-weighted averaging. The probabilistic forecasts produced by the single-model system as well as the multi-model system have been tested against observed rainfall data over 3 austral summer 6-month periods from 2006/07 to 2008/09, using the Brier skill score, relative operating characteristics, and the reliability diagram. The forecast system was found to be more skilful than the persistence forecast. Moreover, the system outscores the forecast skill of the individual models.

show abstract

Deterministic Ensemble Forecasts Using Gene-Expression Programming*

Cited by 39 publications

References 36 publications

Genetic Programming for the Induction of Seasonal Forecasts: A Study on Weather Derivatives

Genetic Programming for the Induction of Seasonal Forecasts: A Study on Weather Derivatives

Electric Load Forecasting for Western Canada: A Comparison of Two Non-Linear Methods

A short-range weather prediction system for South Africa based on a multi-model approach

Contact Info

Product

Resources

About