Softcomputing is a powerful tool that is suited to handle complex problems of modelling and optimisation. The relationships between fibres, production parameters and spinning results is such an example. The relationships can be modelled using neural networks, learning classifier systems and memory based learning. A comparison is made of these techniques with the traditional statistical regression models. For optimisation, price and spinnability are taking into account. It is shown that the price of the cotton blend can be reduced by several percents.