BENCHMARKING AMONG ARTIFICIAL INTELLIGENCE TECHNIQUES APPLIED TO FORECAST

The article is about creating a space for multiple tests of demand forecasting techniques, this space is a software development where besides to testing the algorithms on the same database, these code routines can be compared with each other, this tool allows generate forecasts to be usable in decision making on purchases of Distribution Companies. Besides comparing forecasting some simple techniques like Moving Average (MM) and Last Period with other techniques such as Artificial Neural Networks (ARN) and genetic algorithms (GA), the comparison is made taking into account the error criteria of generated forecasts and the processing time of the methods. Throughout the article explains the design, development and implementation of the above methods and their integration with the tool