[B! ãƒ¡ãƒ¢] silfacã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯

silfac id:silfac

ãƒ¡ãƒ¢ã«é–¢ã™ã‚‹silfacã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ (6)

${{author_name}}$

{{{comment_expanded}}}

{{label}}

{{#is_bookmark}}ãƒªã‚¹ãƒˆ{{/is_bookmark}}{{^is_bookmark}}ãƒªãƒ³ã‚¯{{/is_bookmark}}

${{author_name}}$
{{author_name}}{{created}}
{{ #comment }}{{ comment }}{{ /comment }}
- {{ label }}

${{author_name}}$

{{{comment_expanded}}}

{{label}}

{{#is_bookmark}}ãƒªã‚¹ãƒˆ{{/is_bookmark}}{{^is_bookmark}}ãƒªãƒ³ã‚¯{{/is_bookmark}}

SoftComputing lab.
ãƒ•ã‚¡ã‚¸ã‚£æŽ¨è«– ã‚³ãƒ³ãƒ”ãƒ¥ãƒ¼ã‚¿ã«æŽ¨è«–ã‚’è¡Œã‚ã›ã‚‹ã«ã¯æ§˜ã€…ãªæ‰‹æ³•ãŒã‚ã‚Šã¾ã™ãŒã€ãƒ•ã‚¡ã‚¸ã‚£æŽ¨è«–ã¯ãã®ä¸ã®ä¸€ã¤ã§ã™ã€‚ç‰¹å¾´ã¨ã—ã¦ã¯æŽ¨è«–æ©Ÿæ§‹ã«ãƒ•ã‚¡ã‚¸ã‚£ã®è€ƒãˆæ–¹ã‚’å«ã‚ãŸã“ã¨ã§ã€å‰ææ¡ä»¶ã¨ç”ãˆã¨ã®é–¢ä¿‚ã«å¹…ã‚’æŒãŸã›ã‚‹ã“ã¨ãŒã§ãã¾ã™ã€‚ã—ã‹ã—ã€ãƒ•ã‚¡ã‚¸ã‚£æŽ¨è«–ã¯ã‚ã„ã¾ã„ã•ã‚’æ‰±ã†æŽ¨è«–ãªã®ã§ã€å¸¸ã«ã¯ã£ãã‚Šã¨ã—ãŸï¼ˆã‚¯ãƒªã‚¹ãƒ—ãªï¼‰å†…å®¹ã®æŽ¨è«–ã—ã‹è¡Œã‚ãªã„ã®ã§ã‚ã‚Œã°ã€æ„å‘³ã‚’æˆã—ã¾ã›ã‚“ã€‚ä»¥ä¸‹ã§ã¯ã€ãƒ•ã‚¡ã‚¸ã‚£æŽ¨è«–ã®æ¦‚è¦ã‚’èª¬æ˜Žã™ã‚‹ã“ã¨ã§ã©ã‚“ãªå ´é¢ã«æœ‰ç”¨ã§ã‚ã‚‹ã®ã‹ã¨ã„ã†ã“ã¨ã«ã¤ã„ã¦ã‚‚ç†è§£ã—ã¦ã„ãŸã ã‘ã‚‹ã‚‚ã®ã¨æ€ã„ã¾ã™ã€‚ ãƒ•ã‚¡ã‚¸ã‚£æŽ¨è«–ã¯ã€å‘½é¡Œè«–ç†ã®æ¼”ç®—ã«ã€ã‚ã„ã¾ã„ã•ï¼ˆãƒ•ã‚¡ã‚¸ã‚£ãƒã‚¹ï¼‰ã‚’ç”¨ã„ã‚‹ã“ã¨ã«ã‚ˆã‚Šæ‹¡å¼µã‚’è¡Œã£ãŸæŽ¨è«–æ³•ã¨è¨€ãˆã¾ã™ã€‚ãã“ã§ã€ã¾ãšå‘½é¡Œè«–ç†ã«ã¤ã„ã¦ç°¡å˜ã«èª¬æ˜Žã—ãŸã„ã¨æ€ã„ã¾ã™ã€‚ å‘½é¡Œè«–ç†ã§ã¯ã€ã„ã‚ã‚†ã‚‹ã€Œä¸‰æ®µè«–æ³•ã€ã«ã‚ˆã‚ŠæŽ¨è«–ã‚’è¡Œã„ã¾ã™ã€‚å…·ä½“çš„æ‰‹é †ã¨ã—ã¦ã¯ã€ã¾ãšå‰ææ¡ä»¶ã¨ã—ã¦ã€ã€Œå¸¸è˜çš„çŸ¥è˜ã€ã¨ã€Œç¾åœ¨ã®æ¡ä»¶ã€ã‚’ç”¨æ„ã—ã¾ã™ã€‚ãã—ã¦ã€å¸°ç´çš„ã«ç”ãˆã‚’å°Žã
silfac 2011/04/15
ãƒ•ã‚¡ã‚¸ã‚£æŽ¨è«–ï¼

ãƒ¡ãƒ¢

ãƒ•ã‚¡ã‚¸ã‚£
ãƒªãƒ³ã‚¯
Precision and recall - Wikipedia
Precision and recall In pattern recognition, information retrieval, object detection and classification (machine learning), precision and recall are performance metrics that apply to data retrieved from a collection, corpus or sample space. Precision (also called positive predictive value) is the fraction of relevant instances among the retrieved instances. Written as a formula: Recall (also known
silfac 2011/04/14
Få€¤

ãƒ¡ãƒ¢
ãƒªãƒ³ã‚¯
ã‚¹ãƒ©ã‚¤ãƒ‰ 1
è©•ä¾¡æ–¹æ³• ä¸€èˆ¬çš„ãªãƒ‡ãƒ¼ã‚¿å‡¦ç†çµæžœã®çŠ¶æ…‹ â€¢ å‡¦ç†ï½“ã§çµæžœã®ãƒ‡ãƒ¼ã‚¿é›†åˆãŒå¾—ã‚‰ã‚ŒãŸã€‚ã—ã‹ã—ã€çµæžœã®ä¸ã«ã¯é–“é• ã„ã‚‚ã‚ã‚‹ã—ã€å¾—ã‚‰ã‚Œãªã‹ã£ãŸãƒ‡ãƒ¼ã‚¿ã®ä¸ã«ã‚‚æ£è§£ãŒã‚ã‚Šã†ã‚‹ã€‚ ãƒ‡ãƒ¼ã‚¿é›†åˆå…¨ä½“ å‡¦ç†ï½“ã§å¾—ã‚‰ã‚Œã‚‹ã¹ããƒ‡ãƒ¼ã‚¿é›†åˆ å‡¦ç†ï½“ã§å¾—ã‚‰ã‚ŒãŸ ãƒ‡ãƒ¼ã‚¿é›†åˆ fp tp fn tn æ€§èƒ½è©•ä¾¡å°ºåº¦ â€¢ â€¢ â€¢ å†ç¾çŽ‡ é©åˆçŽ‡ã‚ã‚‹ã„ã¯ç²¾åº¦ ãƒ•ã‚©ãƒ¼ãƒ«ã‚¢ã‚¦ãƒˆ recall = tp tp + fn precision = fallout = tn tn + fp tp tp + fp â€¢ ä¸€èˆ¬æ€§ generarity = tp tp + tn + fp + fn â€¢ Accuracy accuracy = tp + tn tp + fp + fn + tn å†ç¾çŽ‡ ï½–ï½“ é©åˆçŽ‡ â€¢ ã‚ˆãä½¿ã†è©•ä¾¡ã®è¡¨ç¾æ³• ï¼‘ï¼Žï¼ é© åˆ çŽ‡ å†ç¾çŽ‡ï¼‘ï¼ï¼ï¼…ã®è‡ªæ˜Ž ãªã‚·ã‚¹ãƒ†ãƒ ï¼Ÿï¼Ÿ ï¼ï¼Žï¼ ï¼
silfac 2011/04/14
ã“ã‚Œãƒ¡ãƒ¢ã—ãŸã£ã‘ï¼Ÿ

ãƒ¡ãƒ¢
ãƒªãƒ³ã‚¯
Få€¤ã®å‰èº«ã¯Eå€¤? - ã‚·ãƒªã‚³ãƒ³ã®è°·ã®ã‚¾ãƒ³ãƒ“
æœ¬å½“ã¯Få€¤ã§ä½¿ã‚ã‚Œã¦ã„ã‚‹èª¿å’Œå¹³å‡ãŒç›¸åŠ å¹³å‡ï¼Œç›¸ä¹—å¹³å‡ã‚ˆã‚Šã‚‚å°ã•ããªã‚‹ã¨ã„ã†è¨¼æ˜Žã‚’æ›¸ãã¤ã‚‚ã‚Šã ã£ãŸã®ã ã‘ã‚Œã©ï¼ŒFå€¤ã«ã¤ã„ã¦èª¿ã¹ã¦ã„ã‚‹ã†ã¡ã«çŸ¥ã‚‰ãªã‹ã£ãŸæ´å²ãŒå°‘ã—ã ã‘ç´è§£ã‹ã‚ŒãŸï¼Ž Få€¤ã®åˆå‡ºã«ã¤ã„ã¦èª¿ã¹ã‚‹ä½™åŠ›ãŒãªã‹ã£ãŸã®ã§ï¼ŒFå€¤ã®åˆå‡ºã‚’ã”å˜ã˜ã®æ–¹ãŒã„ã‚Œã°ï¼Œæ•™ãˆã¦é ‚ã‘ã‚‹ã¨å¹¸ã„ã§ã™ï¼Ž ã•ã¦Få€¤ã¨ã„ã†è©•ä¾¡æŒ‡æ¨™ã‚’çŸ¥ã‚‰ãªã„æ–¹ã®ãŸã‚ã«ã¡ã‚‡ã£ã¨ã ã‘èª¬æ˜Žï¼Žæ¤œç´¢è©•ä¾¡ã«é™ã‚‰ãšï¼Œæ©Ÿæ¢°å¦ç¿’ãªã©ã®è©•ä¾¡ã§ãŠé¦´æŸ“ã¿ã®precision (é©åˆçŽ‡)ï¼Œrecall (å†ç¾çŽ‡)ã¨ã„ã†ã‚‚ã®ãŒã‚ã‚‹ï¼Ž æ¤œç´¢ã•ã‚ŒãŸæ–‡æ›¸ æ¤œç´¢ã•ã‚Œãªã‹ã£ãŸæ–‡æ›¸ é©åˆ A B ä¸é©åˆ C D ã¨ã™ã‚‹ã¨ï¼Œprecision Pã¯æ¤œç´¢ã•ã‚ŒãŸæ–‡æ›¸ã®ã†ã¡ï¼Œé©åˆæ–‡æ›¸ã®å‰²åˆ ã§è¨ˆç®—ã—ï¼Œrecall Rã¯é©åˆæ–‡æ›¸ãŒã©ã‚Œã ã‘æ¤œç´¢ã•ã‚ŒãŸã®ã‹ã¨ã„ã†å‰²åˆ ã§è¨ˆç®—ã™ã‚‹ï¼Žã•ã¦ï¼Œã¿ã‚“ãªå¤§å¥½ãFå€¤ã¯ï¼Œ ã¨ã—ã¦è¨ˆç®—ã•ã‚Œã‚‹ã“ã¨ãŒå¤šã„ãŒï¼Œã“ã‚Œã¯æ£ç¢ºã«ã¯F1å€¤ã¨å‘¼ã°ã‚Œï¼Œprecisionã¨rec
silfac 2011/04/14
Få€¤ï¼ˆF1å€¤ï¼‰ã®å¤‰å½¢å‰ã‹ã‚‰ã®ã¯ãªã—

ãƒ¡ãƒ¢
ãƒªãƒ³ã‚¯
All SC&I | School of Communication and Information | Rutgers University
silfac 2011/04/14
Recallã¨Precision

ãƒ¡ãƒ¢
ãƒªãƒ³ã‚¯
æ¤œç´¢ã«ãŠã‘ã‚‹é©åˆçŽ‡ (Precision) ã¨å†ç¾çŽ‡ (Recall)
æ¤œç´¢ã«ãŠã‘ã‚‹é©åˆçŽ‡ (Precision) ã¨å†ç¾çŽ‡ (Recall) 2008-01-17-1 [IIR] ã€ŒIntroduction to Information Retrievalã€[1] ã®è¼ªè¬›ã®ç¬¬ä¸€å›ž[2008-01-12-1]ã§ã¡ã‚‰ã£ã¨è©±ã—ã¾ã—ãŸãŒã€ç¬¬ä¸€ç« ã® 1.1 ã« Precision ã¨ Recall ã®èª¬æ˜ŽãŒã‚ã‚Šã¾ã™ï¼ˆç¬¬å…«ç« ã§ã‚‚å‡ºã¦ãã¾ã™ï¼‰ã€‚ è‹¥å¹²æ··ä¹±ã—ã‚„ã™ãã‚„ã‚„ã“ã—ã„è©±ãªã®ã§ã€ã“ã“ã§æ”¹ã‚ã¦è§£èª¬ã—ã¾ã™ã€‚ Â§ Precision (é©åˆçŽ‡) ã¨ã¯ã€ å…¨æ¤œç´¢çµæžœã«å¯¾ã—ã¦ã®ã€ æ¤œç´¢è¦æ±‚ (information need) ã‚’æº€ãŸã™æ¤œç´¢çµæžœã®å‰²åˆã§ã™ã€‚ ä¾‹ãˆã°ã€ ã€ŒMacBook Air ã®é‡é‡ã‚’çŸ¥ã‚ŠãŸã„ã€ã¨ã„ã†æ¤œç´¢è¦æ±‚ã‚’æº€ãŸã™ãŸã‚ã«æ¤œç´¢ã‚ãƒ¼ã€ŒMacBook Air é‡ã•ã€ã§ã‚¦ã‚§ãƒ–æ¤œç´¢ã—ãŸçµæžœ100ä»¶ã®ã†ã¡ã€æ¤œç´¢è¦æ±‚ã‚’æº€ãŸã™ï¼ˆé‡ã•ãŒåˆ†ã‹ã‚‹ï¼‰ã®ãŒ85ä»¶ã ã¨ã™ã‚‹ã¨ã€ Precis
silfac 2011/04/14
Få€¤

ãƒ¡ãƒ¢
ãƒªãƒ³ã‚¯
1