Regression ananlysis?

In a regression ysis about the sales of a product corresponding to the week they were sold, why is it unwise to use the variables 'week' 'week squared' 'log week' and 'inverse week' in the same model?