Resumen
Accurate estimation of the size and development effort for software projects requires estimation models which can be used early enough in the development life cycle to be of practical value. Function Point Analysis (FPA) has become possibly the most widely used estimation technique in practice. However the technique was developed in the data processing environment of the 1970's and, despite undergoing considerable reassessment and formalisation, still attracts criticism for the weighting scoring it employs and for the way in which the function point score is adapted for specific system characteristics. This paper reviews the validity of the weighting scheme and the value of adjusting for system characteristics by studying their effect in a sample of 299 software developments. In general the value adjustment scheme does not appear to cater for differences in productivity. The weighting scheme used to adjust system components in terms of being simple, average or complex also appears suspect and should be redesigned to provide a more realistic estimate of system functionality.