For mission planners and evaluators alike, value in cost models comes from a mean or median prediction, an understanding of the uncertainty on that prediction, and an understanding of model performance. Here we apply advanced statistical and machine learning methods to spacecraft flight software cost, effort, and SLOC estimation, and present the results in the latest version of the Analogy Software Cost Tool (ASCoT). We present in- and out-of-sample performance metrics for our models, each of which incorporate some amount of epistemic uncertainty. ASCoT, hosted on the One NASA Cost Engineering (ONCE) database via the Online NASA Space Estimation Tool (ONSET), was first showcased in 2016 as a number of analogy-based models and methods (k NN and Clustering) to support early project formulation. This ASCoT update improves upon the previous analogic methods by incorporating uncertainty in the data transformations. In particular, we use a Nonlinear Principal Components Analysis (NLPCA) to deal with ordinal data.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    ASCoT 3: Nonlinear Principal Components Analysis and Uncertainty Quantification in Early Concept Spacecraft Flight Software Cost Estimation


    Contributors:


    Publication date :

    2023-03-04


    Size :

    1928630 byte




    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English



    ASCot: the official release a web-based flight software estimation tool

    Johnson, James / Youmans, Tom / Saing, Michael et al. | NTRS | 2017




    Jaguar Bertone XJS Ascot

    Bertone,IT / Jaguar,GB | Automotive engineering | 1977


    Bertone Jaguar XJS Ascot

    Bertone,IT / British Leyland,Jaguar,GB | Automotive engineering | 1977