Weather prediction system is one of the main important applications in agricultural field and it has been technologically and scientifically challenging problem around the world from the past few decades. Many of the crops depend upon ideal weather condition for example tea plants are very sensitive to sun light. If the tea plants are exposed to direct sun light or temperature beyond 32 degree centigrade the tea plants get die. In the same way if there is huge rain fall before Jawar crop, the crop might get damaged, similarly if the humidity is extremely high the cotton crop gets damaged, but we focus on only temperature, by seeing the temperature value only the farmers know which crop would be suitable for which weather condition. If a former knows that next year temperature will be quite high during the time of a particular crop, farmers might opt for another crop, without knowing that if farmer goes for same crop that crop might be damaged. In this project we will overcome from that problem, weather prediction system is very important to help the farmer to analysis before  plan the seeds to know which crop would be better for the specific weather, not only the weather but also what type of soil is suitable for which crops is also considered. Data mining techniques can be used for predicting the weather forecast.

In this paper we examine the use of data mining techniques for predicting the temperature, rainfall, humidity, wind speed, visibility and pressure. This can be obtained by using Multilayer Artificial Neural Network and also we use Median Filtering Techniques and we have to take the previous year weather data from 2005 to 2015 from agricultural department. A Predictive Neural Network model is developed for the weather prediction and the result is compared with real data. This information is helpful for the farmers before seeding the crop to get good yield out of particular crop.