The modelling and quantification of materials data using numerically optimised mathematical functions is an important and widespread practice. In the present paper the development of robust and exploratory data analysis strategies, based on exploratory data analysis procedures and robust statistics, is described. Application of this new PC based data analysis system to two sets of established data which relate materials information to process variables demonstrates the advantages of a comprehensive and robust approach to analysis. The advantages of techniques that use both ‘degree off it’ and also predictive statistical criteria for mathematical/statistical models are described.
MST/1945
Get full access to this article
View all access options for this article.
References
1.
NORRISHJ. and STRUTTJ. E.: Weld. Met. Fabr., October 1988, 56, (7), 337–341.
2.
CARYH. B.: Weld. J.,1991, 70, 40–45.
3.
SIEWARTT. A.: Weld. T.,1986, 65, (3), 23–28.
4.
ALBERRYP. J., BRUNNSTROMR. R. L., and JONESK. E.: Met. Technol., January 1983, 10, 28–38.
5.
PRESSW. H., FLANNERYB. P., TEUKOLSKYS. A., and VETTERLINGW. T.: ‘Numerical recipes’, 1986, Cambridge, Cambridge University Press.
6.
HOAGLIND. C., MOSTELLERF., and TUKEYJ. W.: ‘Understanding robust and exploratory data analysis'; 1983, New York, Wiley.
7.
ALLEND. M.: Technotnetrics, February 1974, 1, (16), 125–127.
