Rain models
Rain precipitation is crucial for different people, cultures, organizations and governments around the world. Reliable rain models do exist, but different requirements in terms predicting precipitation duration and amount are often required. Our client, based in Malaysia, required an extensive predictive model based on historical data and ‘smart’ forecasts. Malaysian economy is heavily dependent on rain and its accurate and reliable predictions.
Malaysian rain records for the last 30 years were provided by our client and existing models’ predictive failures were analyzed and investigated. Our advantages included access to neural network building software, machine learning approaches and other predictive methodologies. We utilized similar predictive failures from other Asian regions with similar characteristics and ‘fed’ them to our artificial neural networks. The outcome was a trained artificial neural network with higher predictive capacity both in terms of duration and weight of precipitation. Attributes of this artificial neural network cannot be shared as they are property of our client.