平成20（2008）年度　共同利用登録実施報告書

課題番号	20−共研−10	分野分類		統計数理研究所内分野分類		a
				主要研究分野分類		7
研究課題名	貧困問題の空間解析と統計的モデリング
フリガナ 代表者氏名	ﾊｯｻﾌﾞ　ｴﾙｶﾞｳｨ　ｵｽﾏﾝ		ローマ字		HASSAB ELGAWI OSMAN
所属機関	東京大学
所属部局	新領域創成科学研究科
職　　名	博士（.Ph.D student)

研究目的と成果の概要

My current research work always challenges with statistical estimations and modeling. That why I undertook the statistical aspect of my research project at the ISM to get advices from Professor Ogata for the relevant methodologies and validations, especially for the spatial or space-time analysis.
First, I studied the effect of spatial country characteristics on economic growth and development. Though statistically weak there is evidence about the correlation between location (spatial determinants) and underdevelopment. The focus is to envision on spatial econometric validity regard such assumption with alternative conceptions of space other than geographical distance. Then I studied statistical learning theory to overcome some important limitations often encountered on conventional statistical regression techniques such as least square (LS), semi-parametric and non-parametric bayesian models when applied to a real system e.g. complex economic system. In addition to the tradeoff between higher prediction variance (due to using too many regressors) and biased prediction (due to using too few regressor variables), the curse of dimensionality imposes another limitation and that some of these can be overcome by incorporating complementary learning process and variable selection mechanism.
A considerable amount of all statistical models I developed there significantly contributed to my doctoral dissertation and towards the final goal to develop a comprehensive assessment of factors hampering growth and development based on spatial statistical analysis.