統計的因果推論勉強会 第2回

経営学系統計学エンドユーザーのための，統計的因果推論勉強会の第2回です。ブログエントリーはこちらです。
http://hikaru1122.hatenadiary.jp/entry/2016/06/25/221339

Transcript

1. ౷ܭతҼՌਪ࿦
   ษڧձ
   ୈ2ճ
   2016೥6݄25೔
   1
   

   View Slide

2. ౰ษڧձͷ࢟੎
   • จܥܦӦֶܥ౷ܭֶΤϯυϢʔβʔʹΑΔ
   • ΤϯυϢʔβʔͷͨΊͷษڧձ
   • ࣗ෼ͷݚڀʹͲ͏Ԡ༻͢Δ͔Λॏࢹ͢Δ
   2
   

   View Slide

3. ؔ࿈χϡʔε
   ؠ೾σʔλαΠΤϯε
   Vol.3ʰҼՌਪ࿦ʱ͕ൃץʂ
   ͥͻखʹऔͬͯΈ͍ͯͩ͘͞ʂ
   ͍ͬͯ͏͔ɼങ͏ͱ͍͍Ͱ͢ɻ
   3
   

   View Slide

4. ࠓճͷൣғ
   • ٶ઒ຊʰ౷ܭతҼՌਪ࿦ʱୈ2ষɹ17ʙ35ϖʔδ
   • ੕໺ຊʰௐࠪ؍࡯σʔλͷ౷ܭՊֶʱୈ2ষɹ
   35ʙ50ϖʔδ
   • ಋೖ෦ͷ࠷ޙ
   ·࣮ͩࡍͷ෼ੳͰ͖ΔΑ͏ʹ͸ͳΒͳ͍ʂ
   4
   

   View Slide

5. લճͷ෮श
   • ௐࠪ؍࡯ݚڀΛ͍͔ʹ࣮ݧݚڀʹ͚ۙͮΔ͔͕େ
   ࣄɻ
   • γϯϓιϯͷύϥυοΫε
   • ॲஔ܈ʢ࣮ݧ܈ʣͱରর܈ʢඇॲஔ܈ʣ
   • આ໌ม਺ʢಠཱม਺ʣͱ݁Ռม਺ʢैଐม਺ʣ
   5
   

   View Slide

6. ࠓ೔ͷΩʔϫʔυ
   • જࡏ൓ԠϞσϧɾ൓࣮Ծ૝Ϟσϧ
   • ҼՌޮՌɾฏۉॲஔޮՌʢATEʣ
   • ަབྷ
   • ڞมྔ
   • ܏޲είΞ
   6
   

   View Slide

7. ٶ઒ຊɹୈ2ষ
   ʮ౷ܭతҼՌਪ࿦ͷओ໨త͸ɼॲཧม਺͕൓Ԡม਺
   ʹٴ΅͢ҼՌؔ܎ΛఆྔతʹධՁ͠ɼͦΕΛར༻͢
   Δ͜ͱʯʢ17ทʣ
   7
   

   View Slide

8. ٶ઒ຊɹୈ2ষ
   ʮڞ෼ࢄ෼ੳʹ͓͚Δิॿଌఆ஋ͱ͸ɼ࣮ݧͷॳظ
   ஈ֊Ͱͷ͹Β͖ͭΛࣔ͢ڞมྔͰ͋ͬͯɼҼࢠʢॲ
   ཧʣͷӨڹΛड͚ΔதؒಛੑͰ͋ͬͯ͸ͳΒͳ͍ʯ
   ʢ19ทʣ
   ্ͷ ໊͕ٛม਺ʹͳͬͯΔͷ͕ڞ෼ࢄ෼ੳ
   ʢ௒େͬ͟ͺʣ
   8
   

   View Slide

9. ٶ઒ຊɹୈ2ষ
   • ҼՌਪ࿦ͷجຊ໰୊ʢࠜຊ໰୊ʣ
   ʮҼՌޮՌ͸ਪఆͰ͖ͳ͍ʯʢ26ทʣ
   9
   

   View Slide

10. ٶ઒ຊɹୈ2ষ
    • ڧ͍ҙຯͰͷແࢹՄೳੑʢٶ઒ຊʣ
    • ڧ͘ແࢹͰ͖ΔׂΓ౰ͯ৚݅ʢ੕໺ຊʣ
    ʮׂΓ౰ͯ͸͋͘·ͰڞมྔʹͷΈґଘ͠ɼ݁Ռม
    ਺ʹ͸ґଘ͠ͳ͍ʯʢ43ทʣ
    • ແ࡞ҝׂΓ෇͚ͬΆ͘͢Δ
    • Ͳ͏͍͏ͱ͖ʹڧ͍ҙຯͰແࢹՄೳͰ͋Δ͔͸౷
    ܭ෼ੳ͚ͩͰ͸Θ͔Βͳ͍ɻ
    10
    

    View Slide

11. ҼՌμΠΞάϥϜΛ
    ࢖͑ʂ
    11
    

    View Slide

12. ੕໺ຊɹୈ2ষ
    • ൓࣮Ծ૝Ϟσϧʗજࡏ൓ԠϞ
    σϧ
    counterfactual model
    potential outcomes
    model
    • ΋͠΋ͷ͓࿩ɻ྆ํͱ΋஋͕
    ଘࡏ͢Δ͸ͣɻͰ΋ɼ࣮ࡍ͸
    Ұํ͔͠؍ଌ͞Εͳ͍
    • ౷ܭతҼՌਪ࿦͸ܽଌͷσʔ
    λΛ෼ੳ͢Δ͜ͱ
    12
    

    View Slide

13. ੕໺ຊɹୈ2ষ
    • ҼՌޮՌ ʹ
    ʮॲஔ܈ͷ݁Ռʯʔʮରর܈ͷ݁Ռʯͷࠩ
    13
    

    View Slide

14. ͨͩ͠
    ແ࡞ҝׂΓ౰ͯ
    ͩͬͨΒ
    14
    

    View Slide

15. ੕໺ຊɹୈ2ষ
    ʮҼՌޮՌ͸ॲஔ܈ͱରর܈͕ͲͷΑ͏ͳ฼ूஂ͔
    Βநग़͞Εͨ΋ͷͰ͋Δ͔ʹґଘ͢Δʯʢ39ทʣ
    ʮڞมྔͷӨڹΛআڈ͢Δʯʢ41ʙ42ทʣ
    15
    

    View Slide

16. ੕໺ຊɹୈ2ষ
    ʮڞมྔΛ৚݅෇͚Ε͹ɼｙ ,ｙ ͷಉ࣌෼෍ͷܗ
    ͸ͲͪΒͷ܈ʹ෼͚ΒΕ͔ͨʹ͸ґଘ͠ͳ͍ʯʢ44
    ทʣ
    16
    

    View Slide

17. ੕໺ຊɹୈ2ষ
    • ͡Ό͋ɼڞมྔΛͲ͏ௐ੔͢Δ͔ʁ
    • طଘͷҼՌޮՌͷਪఆํ๏ʹ͸͍Ζ͍Ζ໰୊͕͋
    Δʢ45ʙ57ทʣ
    17
    

    View Slide

18. ͦ͜Ͱɼηϛύϥ
    ϝτϦοΫͳਪఆ
    Λߦ͓͏ʂ
    18
    

    View Slide

19. ࣍ճʹͭͮ͘ʢຊ͸ʣ
    19
    

    View Slide

20. ܏޲είΞΛ࢖ͬͨ؆୯ͳ෼ੳྫ
    ΍ͬͯΈΑ͏ʂ
    • ੴଜɾੴଜʢ2015ʣͷσʔλΛ͓आΓ͠·͠
    ͨɻͦͷࡍɼঢ়گઃఆΛม͑·ͨ͠ɻ
    ͢Έ·ͤΜɻ
    • ঢ়گม͍͑ͯΔͷͰɼ݁Ռ͸σλϥϝͰ͢ɻ
    • ໰ʮঢਐ͸޾෱౓ʹରͯ͠ҼՌޮՌΛ͍࣋ͬͯΔ
    ͔ʯ
    20
    

    View Slide

21. View Slide

22. σʔλʹ͍ͭͯ
    • ڞมྔ͸ੑผɼίϛϡྗɼചΓ্͛
    22
    

    View Slide

23. ී௨ʹtݕఆʢ༗ҙࠩͳ͠ʣ
    #ͱΓ͋͑ͣ܈ผͰฏۉΛݟͯΈΔɻ
    > by(dat[,"happy"], dat[,"promote"], mean)
    dat[, "promote"]: ঢਐͳ͠
    [1] 2.853333
    ------------
    dat[, "promote"]: ঢਐ͋Γ
    [1] 4.48
    #tݕఆ
    > t.test(promote0$happy, promote1$happy)
    Welch Two Sample t-test
    data: promote0$happy and promote1$happy
    t = -2.2515, df = 6.6525, p-value = 0.06101
    alternative hypothesis: true difference in means is not equal to 0
    95 percent confidence interval:
    -3.35328611 0.09995278
    sample estimates:
    mean of x mean of y
    2.853333 4.480000
    23
    

    View Slide

24. ܏޲είΞͰϚονϯά
    #܏޲είΞΛग़͢ɻ
    ps_score <- glm(promote ~ gender + commu + sales, family = binomial, data = dat2)
    #ϚονϯάʹඞཁͳσʔλΛऔΓग़͢ɻ
    dat$keikou.score <- ps_score$fitted.values
    keikou.score <- ps_score$fitted.values
    happy <- dat2$happy
    warituke <- dat2$promote
    #ϚονϯάͰ2܈ͷฏۉͷࠩΛݟΔɻ
    kekka <- Match(Y=happy, Tr=warituke, X=keikou.score, M=1)
    summary(kekka)
    • ATE Ͱ͸ͳ͘ ATT(TET) Ͱ͢ɻ
    24
    

    View Slide

25. ༗ҙͳ͕ࠩݟΒΕΔɻ
    > summary(kekka)
    Estimate... 2.2
    AI SE...... 0.90173
    T-stat..... 2.4398
    p.val...... 0.014697
    Original number of observations.............. 20
    Original number of treated obs............... 5
    Matched number of observations............... 5
    Matched number of observations (unweighted). 5
    • ঢਐʹΑͬͯ޾෱౓͕ߴ·Δʢ΢ιʣ
    25
    

    View Slide

26. ͲΜͳ૊Έ߹Θ͔ͤͩͬͨʁ
    • ͔ͨ͠ʹ܏޲είΞ͕͍ۙϚονϯά
    > kekka$index.treated
    [1] 16 17 18 19 20
    > kekka$index.control
    [1] 4 6 12 3 12
    26
    

    View Slide

27. ࢀߟจݙ
    ٶ઒ຊ
    ੕໺ຊ
    ੴଜః෉ɾੴଜ༑ೋ࿠ʢ2015ʣʮSPSSʹΑΔ܏޲
    είΞͱϚονϯάͷखॱʯɼʰ௽ݟେֶلཁɹୈ
    4෦, ਓจɾࣾձɾࣗવՊֶฤʱ(52), 31-34ɻ
    27
    

    View Slide