The outcomes make sure that our model has the potential for effective application for weather prediction.
Components of a modern weather forecasting system
include the following modules:
• Data Collection
• Data Pre-processing
• Training
• Testing
• Results/Output
1.Data collection:
In this step numerous sensors are used to collect the data like wind sensors, rain sensor, pressure sensor and temperature sensors. The data is collected repeatedly when short interval of time so we’ve enough input for the process. Large dataset helps in increasing the accuracy of the output. After this the data is send for pre-processing.
2.Data pre-processing:
The Pre-processing step is employed to remove the unwanted data or noise recorded by the sensors during transmission or it may refer to the selection of a pecific area for consideration for prediction purpose.
3.Training:
Training of the network is done with the help of back propagation algorithm. During this we’ve to choose learning rate and therefore the momentum value. Learning rate control the speed of the network. As we have a tendency to increase the learning rate it’ll speed up the training. We are using tan h activation function as a result of it provides more recognition and accuracy. We can use any number of hidden layers or can increase it if the network is not learning well. The input of the back propagation network is temperature, rainfall, humidity, pressure and precipitation.
4.Testing:
Testing is completed by providing with numerous different dataset as input and obtaining the specified result. When we don’t get correct results than the network is once more trained. Network is checked with completely different situations which will occur in future.