Calibration and Validation of the Cocomo II.1997.0 Cost/Schedulee Estimating Model to the Space and Missile Systems Center Database
Versandkostenfrei!
Versandfertig in über 4 Wochen
53,99 €
inkl. MwSt.
PAYBACK Punkte
27 °P sammeln!
The pressure to decrease costs within the Department of Defense has influenced the start of many cost estimating studies, in an effort to provide more accurate estimating and reduce costs. The goal of this study was to determine the accuracy of COCOMO II.1997.0, a software cost and schedule estimating model, using Magnitude of Relative Error, Mean Magnitude of Relative Error, Relative Root Mean Square, and a 25 percent Prediction Level. Effort estimates were completed using the model in default and in calibrated mode. Calibration was accomplished by dividing four stratified data sets into two ...
The pressure to decrease costs within the Department of Defense has influenced the start of many cost estimating studies, in an effort to provide more accurate estimating and reduce costs. The goal of this study was to determine the accuracy of COCOMO II.1997.0, a software cost and schedule estimating model, using Magnitude of Relative Error, Mean Magnitude of Relative Error, Relative Root Mean Square, and a 25 percent Prediction Level. Effort estimates were completed using the model in default and in calibrated mode. Calibration was accomplished by dividing four stratified data sets into two random validation and calibration data sets using five times resampling. The accuracy results were poor; the best having an accuracy of only .3332 within 40 percent of the time in calibrated mode. It was found that homogeneous data is the key to producing the best results, and the model typically underestimates. The second part of this thesis was to try and improve upon the default mode estimates. This was accomplished by regressing the model estimates to the actual effort. Each original regression equation was transformed and tested for normality, equal variance, and significance. Overall, the results were promising; regression improved the accuracy in three of the four cases, the best having an accuracy of .2059 within 75 percent of the time.
Andere Kunden interessierten sich für
