Chapter 1 회귀분석 - wolfpack.hnu.ac.kr

1 WolfpackREGRESSION / 1 . 1 Prof. Sehyug Kwon, Dept. of Statistics, HANNAM University @2005 Spring (regress) go back to an earlier and worse condition ( ) . Francis Galton(1822-1911) . Galton ( sweat pea) 928 ( ) ( ) .. Galton . Galton Karl Pearson(1903) . Pearson 1078 . ) ( ) (XY+= Chapter 1 wolfpackREGRESSION / 1.

2 2 Prof. Sehyug Kwon, Dept. of Statistics, HANNAM University @2005 Spring . (statistical hypothesis) (model) (significance) . Confirmatory( ) .. Tukey EDA, Data Mining .. , , , . (data trail, data history) ( , , , , , ) . (variable), (subject) . ( ) (observation) . ( ).

3 (discrete): . , , , . (continuous): (infinite) . (range) , , IQ, . (metric, measurable, quantitative): , , , IQ, , .. ) , ( ) wolfpackREGRESSION / 1 . 3 Prof. Sehyug Kwon, Dept. of Statistics, HANNAM University @2005 Spring ( ) (non-metric, classified, categorical, qualitative): , . (1) (nominal): . , , (2) (ordinal): . (A, B, ..) ( , , ), (5 ) (time series) (cross-section: ).

4 ( , ) ( ), . , (econometrics) ( ) . (casual relationship) ( ) (exploratory variable) (independent) (dependent) (response) . Y, X . , .. eXXXfYp+=),..,,(21 , .. 40 . Y, X (significant) ? (statistical software) , (statistic) ( , , ).

5 SPSS(Statistical wolfpackREGRESSION / 1 . 4 Prof. Sehyug Kwon, Dept. of Statistics, HANNAM University @2005 Spring Package for Social Science), QC Minitab, SYSSTAT, STATGRAPHICS, S-plus, RATS, EVIEW . EXCEL (add-on ) . , , . SAS North Carolina Raleigh NCSU(North Carolina State University) Statistical Analysis System 1966 . 1972 SAS72 Shareware , 1976 Cary (NCSU 15 ) SAS Institute SAS.

6 (SAS : Strategic Application System) . SAS 118 ( , 57 ), 40,000 ( , , , ) , Fortune 500 90% SAS . version 8, version 9 . SPSS Inc. 1968 (raw data) SPSS 12 . SAS .. SPSS *.sav *.spo .. , ( =1, =0) .. (parameter ) ),..,,(21nxxx wolfpackREGRESSION / 1 . 5 Prof.

7 Sehyug Kwon, Dept. of Statistics, HANNAM University @2005 Spring (statistics: , , , ) ( ) . (estimator) (test statistic) . (point estimation) (interval estimation) .. why? . (95% , 5% , ) (nonparametric, distribution free) .. (stem and leaf) (box-whisker) (histogram) : ( , ) , , , IQR ( , ), (*) t- (*) (Sign test, Wilcoxon rank sum test) (*) 2 - (bar) , (pie) (proportion, ratio) ( ).

8 (*) z- ( ) (*) (p-value) ( ) (*) z- ( ) ( ) . (*) z- ( ) (*) (p-value) ( ) (t- ) (z- ) , (F- ), (t- ) . 0 .. (box-whisker) : ( , ) , , , IQR wolfpackREGRESSION / 1 . 6 Prof. Sehyug Kwon, Dept. of Statistics, HANNAM University @2005 Spring , (*) t- (*) F- (bar) (proportion, ratio) (*) z- ( ) (*) ( ) 00: =H : )1,0( )1(~/0 NorntnsxT = : nsntx)2/1;1( 00:ppH= : )1,0(~ 000 NnqpppT =( ).

9 : nqpzp 2/1 210: =H :)1,0( )2(~110)(212121 NorntnnsxxTp += = 2)1()1(21222211 + + =nnsnsnsp : 212111)2/1;2(nnsntxxp+ 210:ppH= wolfpackREGRESSION / 1 . 7 Prof. Sehyug Kwon, Dept. of Statistics, HANNAM University @2005 Spring : )1,0(~ 0)( 2221112121 NnqpnqpppppT+= = ( ) EXACT test : 2221112/121 nqpnqpzppT+ = wolfpackREGRESSION / 1 . 8 Prof. Sehyug Kwon, Dept. of Statistics, HANNAM University @2005 Spring ( 2 ) 1) (casual relationship) , ( : regression, ( ) : (Multivariate) ANOVA) 2) (correlation) (reduction) (classification).

10 ( + ) (*) : ( ) Logit ( ) (2 - ) : Log-normal . ( ) ( ) ( ) ( , , ) . EXAMPLE 1-1 : EXAMPLE1-3 : 145 . 145 ? 5% EXAMPLE 1-2 : ( ) .. : 145 150 153 148 141 152 146 154 139 148 : 152 150 147 155 140 146 158 152 151 143 wolfpackREGRESSION / 1.