Dengue fever is a systemic viral infection of epidemic proportions in tropical countries. The incidence of dengue fever is ever increasing and has doubled over the last few decades. Estimated 50million new cases are detected each year and close to 10000 deaths occur each year. Epidemics are unpredictable and unprecedented. When epidemics occur, health services are over whelmed leading to overcrowding of hospitals. At present there is no evidence that dengue epidemics can be predicted. Since the breeding of the dengue mosquito is directly influenced by environmental factors, it is plausible that epidemics could be predicted using weather data. We hypothesized that there is a mathematical relationship between incidence of dengue fever and environmental factors and if such relationship exists, new cases of dengue fever in the succeeding months can be predicted using weather data of the current month. We developed a mathematical model using machine learning technique. We used Island wide dengue epidemiology data, weather data and population density in developing the model. We used incidence of dengue fever, average rain fall, humidity, wind speed, temperature and population density of each district in the model. We found that the model is able to predict the incidence of dengue fever of a given month in a given district with precision (RMSE between 18- 35.3). Further, using weather data of a given month, the number of cases of dengue in succeeding months too can be predicted with precision (RMSE 10.4-30). Health authorities can use existing weather data in predicting epidemics in the immediate future and therefore measures to prevent new cases can be taken and more importantly the authorities can prepare local authorities for outbreaks.