A method and system for forecasting using pattern recognition and extension software. Models of the present invention select patterns from a library that match historical data and extend them into the future to make forecasts that can be used with a variety of predictive technologies.
