مشروع البحث: Using Machine Learning Techniques for Weather Forecasting in the City of Zintan
| dc.contributor.advisor | د.ابوعجيلة المصري دغمان | |
| dc.date.accessioned | 2026-06-07T09:58:47Z | |
| dc.date.available | 2026-06-07T09:58:47Z | |
| dc.description | This study aims to develop the current manual system based on paper records at the Zintan Meteorological Station and transform it into a digital electronic system capable of predicting future weather conditions based on historical data, using machine learning techniques via the WEKA software, "The dataset utilized in this study was acquired from the British website (OpenWeather.com), which is recognized as a highly reliable and credible digital source for meteorological data. The study analyzed weather data from the past ten years (from January 1, 2014, to December 31, 2024), which included (122,461 records) and (9 attributes): temperature, dew point, minimum temperature, maximum temperature, pressure, humidity, wind speed, total cloud cover, and main weather condition, recorded daily with hourly precision. Two testing methods were compared: 10-Folds Cross Validation and Percentage Split (66% training and 34% testing). The 10-Folds Cross Validation method was found to be superior, achieving an accuracy of 81.56% compared to 81.40% for the Percentage Split method. Five machine learning algorithms—Bayes Network, Hoeffding Tree, J48, Random Forest, and Random Tree—were then applied to build models and compare their performance in predicting the nine weather factors. The results indicated that the Random Forest algorithm performed the best, achieving an accuracy of 86.02% with the lowest RMSE of 0.157. The highest prediction accuracy was observed for the attribute Temp_min at 97.34% with the lowest RMSE of 0.026, while the attribute weather_main had the lowest accuracy at 86.02% with the highest RMSE of 0.157. These findings demonstrate that the optimal testing method is 10-Folds Cross Validation and the most suitable algorithm for the dataset is Random Forest, which is recommended for predicting future weather conditions in Zintan. Keywords: Weather prediction, machine learning, WEKA software, Random Forest, 10-Folds Cross Validation, historical weather data, meteorological station, Zintan. | |
| dc.description.abstract | This study aims to develop the current manual system based on paper records at the Zintan Meteorological Station and transform it into a digital electronic system capable of predicting future weather conditions based on historical data, using machine learning techniques via the WEKA software, "The dataset utilized in this study was acquired from the British website (OpenWeather.com), which is recognized as a highly reliable and credible digital source for meteorological data. The study analyzed weather data from the past ten years (from January 1, 2014, to December 31, 2024), which included (122,461 records) and (9 attributes): temperature, dew point, minimum temperature, maximum temperature, pressure, humidity, wind speed, total cloud cover, and main weather condition, recorded daily with hourly precision. Two testing methods were compared: 10-Folds Cross Validation and Percentage Split (66% training and 34% testing). The 10-Folds Cross Validation method was found to be superior, achieving an accuracy of 81.56% compared to 81.40% for the Percentage Split method. Five machine learning algorithms—Bayes Network, Hoeffding Tree, J48, Random Forest, and Random Tree—were then applied to build models and compare their performance in predicting the nine weather factors. The results indicated that the Random Forest algorithm performed the best, achieving an accuracy of 86.02% with the lowest RMSE of 0.157. The highest prediction accuracy was observed for the attribute Temp_min at 97.34% with the lowest RMSE of 0.026, while the attribute weather_main had the lowest accuracy at 86.02% with the highest RMSE of 0.157. These findings demonstrate that the optimal testing method is 10-Folds Cross Validation and the most suitable algorithm for the dataset is Random Forest, which is recommended for predicting future weather conditions in Zintan. Keywords: Weather prediction, machine learning, WEKA software, Random Forest, 10-Folds Cross Validation, historical weather data, meteorological station, Zintan. | |
| dc.identifier | 2-16 | |
| dc.identifier.uri | https://dspace.academy.edu.ly/handle/123456789/2177 | |
| dc.subject | هندسة كهربائية | |
| dc.title | Using Machine Learning Techniques for Weather Forecasting in the City of Zintan | |
| dspace.entity.type | Project | |
| project.endDate | 2026 | |
| project.funder.name | التطبيقية والهندسية | |
| project.investigator | العزيزي الكيلانى أحمد | |
| project.startDate | 2025 |