Implementation of bayesian model averaging on the weather data forecasting applications utilizing open weather map

• Published in: IOP conference series. Materials Science and Engineering 2018-02, Vol.309 (1), p.12054
• Main Authors: Rahmat, R F, Nasution, F R, Seniman, Syahputra, M F, Sitompul, O S
• Format: Article
• Language: English
• Subjects: Accuracy; Bayesian analysis; Economic forecasting; Errors; Forecasting; Forecasting techniques; Humidity; Mathematical models; Meteorological data; Time measurement; Weather forecasting; Wind speed
• ISSN: 1757-8981; 1757-899X
• DOI: 10.1088/1757-899X/309/1/012054

Weather is condition of air in a certain region at a relatively short period of time, measured with various parameters such as; temperature, air preasure, wind velocity, humidity and another phenomenons in the atmosphere. In fact, extreme weather due to global warming would lead to drought, flood, hurricane and other forms of weather occasion, which directly affects social andeconomic activities. Hence, a forecasting technique is to predict weather with distinctive output, particullary mapping process based on GIS with information about current weather status in certain cordinates of each region with capability to forecast for seven days afterward. Data used in this research are retrieved in real time from the server openweathermap and BMKG. In order to obtain a low error rate and high accuracy of forecasting, the authors use Bayesian Model Averaging (BMA) method. The result shows that the BMA method has good accuracy. Forecasting error value is calculated by mean square error shows (MSE). The error value emerges at minumum temperature rated at 0.28 and maximum temperature rated at 0.15. Meanwhile, the error value of minimum humidity rates at 0.38 and the error value of maximum humidity rates at 0.04. Afterall, the forecasting error rate of wind speed is at 0.076. The lower the forecasting error rate, the more optimized the accuracy is.