Prof Yasushi Yoshida presented a Seminar at the Wessex Institute of Technology on "Multi-Variable Analyses Regression by the Least Square and Classification by the Principal Component Analysis".
Yasushi graduated as a Mechanical Engineer at the Japan Defence College, where he also obtained a Master in Civil Engineering. His PhD was awarded by Yokohama University. He started his research career at Japanese Defence Agency where he worked at the Technical Research and Development Institute. More recently, he was appointed Professor at Tsuzuki Institute.
Prof Yoshida has numerous publications in his field of expertise, ie ship sciences, as well as having received some important prizes. He is a member of several research and professional associations and a Fellow of Wessex Institute where he has spent the last year, carrying out research in hydrodynamics.
Yasushi began his presentation by defining the different types of analysis used in applied statistics, ie parametric and non-parametric analysis. With parametric analysis we can have region, discriminative function, principal component and selection analyses. Non-parametric analyses are based on survey plans usually carried out with a questionnaire.
The principal component analysis has been applied by Professor Yoshida to predict the resistance performance of high speed craft models (see page 187 of "Computational Stochastic Mechanics", edited by PD Spanos and CA Brebbia, Computational Mechanics Publications, Southampton and Boston, 1991). The empirical formulae obtained by Prof Yoshida were verified against tow tank data and good argument was found for low to high speed range. The empirical formula developed by Yasushi was based on data obtained from 78 tow tank tests. Since then, the formula has been used to predict the behaviour of new models, with excellent results.