Kendallé ä½ç¸é¢ä¿æ° (Kendall's rank correlation)â \(n\)åã®ãã¼ã¿ã®å¯¾ \((x_1,y_1),\ldots,(x_n,y_n)\) ã®ç¸é¢ã測ãå ´åï¼ n(n-1)/2 åã®å¯¾ \((x_i,y_i)\) 㨠\((x_j,y_j)\) ã«ã¤ãã¦ï¼ åé åºã®å¯¾ã®æ°ã Pï¼éé åºã®å¯¾ã®æ°ã Q ã¨ããï¼ ä¾ãã°ï¼\(x_i\gt y_i\) ã㤠\(x_j\gt y_j\) ãªãåé åºï¼ åé åºããªãå ´åï¼ \[\tau=\frac{P-Q}{n(n-1)/2}\] ãKendallé ä½ç¸é¢ä¿æ°ï¼Kendall Ïã¨ãããï¼ åé ä½ãããå ´åï¼ \(x_i\) ã®ä¸ã§åé ä½ãããåãããã¯ã®å¤§ããã \(t_{xi}\) ã¨ããï¼ä¾ãã°ï¼2ã¨3ä½ï¼4,5,6ä½ãåé ä½ãªãï¼\(t_{x1}=2\) 㨠\(t_{x2}=3\) ã¨ãªãï¼ãã®ã¨ãï¼
è¨ç®æé ï¼ ã±ã¼ã¹æ°ã $n$ ã¨ããã å¤æ° $X$ ã¨å¤æ° $Y$ ã«ã¤ãã¦å°ããæ¹ããé ä½ãã¤ãï¼å¤æ° $X$ ã«ã¤ãã¦å°ããé ã«ä¸¦ã¹å¤ããï¼åé ä½ã®å ´åã«ã¯å¹³åé ä½ãã¤ããï¼ã $Y_{i}\ (i = 1, 2, \dots , n - 1)$ ã«ã¤ãã¦ï¼$Y_{i} \lt Y_{j}$ ã®åæ°ã $P_{i}$ ï¼$Y_{i} \gt Y_{j}$ ã®åæ°ã $Q_{i}$ ã¨ãã$(j = i + 1, i + 2, \dots , n)$ã ä¾ãã°ï¼è¡¨ 2 ã«ç¤ºãããã«ï¼$X_{5}$ ã«å¯¾ãã $Y$ ã®é ä½ã¯ $7$ ã§ããï¼ããããå³ã«ãã $Y$ ã®é ä½ã®ãã¡ï¼å¤§ãããã®ã¯ $Y_{8}, Y_{9}, Y_{10}$ ã® $3$ åï¼$P_{5} = 3$ï¼ï¼å°ãããã®ã¯ $Y_{6}, Y_{7}$ ã® $2$ åï¼$Q_{5} = 2$ï¼ã $P_{i} +
ãç¥ãã
é害
ã©ã³ãã³ã°
ãªãªã¼ã¹ãé害æ å ±ãªã©ã®ãµã¼ãã¹ã®ãç¥ãã
ææ°ã®äººæ°ã¨ã³ããªã¼ã®é ä¿¡
å¦çãå®è¡ä¸ã§ã
j次ã®ããã¯ãã¼ã¯
kåã®ããã¯ãã¼ã¯
lãã¨ã§èªã
eã³ã¡ã³ãä¸è¦§ãéã
oãã¼ã¸ãéã
{{#tags}}- {{label}}
{{/tags}}