ãWhite Noiseð€ãæç³»åãã¯ãçµæžåæã®åºç€æŠå¿µïŒèšéçµæžåŠâšNo.2
IntroductionïŒèšéçµæžåŠãžã®ææŠð¥
çµæžåŠéšã«éãç§ã
ãããã倧åŠãåŠéšãæçµå¹Žã«ãªã
åŠåã«å
šåã泚ãæéãéãããŠããŸããð
ãç¥ã¯åãªãããšããèšèãä¿¡ããŠ
æ®ãã®å€§åŠçç掻ãæºå«ããããšæããŸã
åŠéšã¬ãã«ã®ãã¯ãçµæžåŠã¯
å人çã«ããç解ã§ããã€ããã§ã
ããããªãããæ¬åœã®çµæžã®ååãç解ããã«ã¯ãåŠéšã¬ãã«ã®ç¥èã§ã¯ã話ã«ãªããŸããð¥
ãŸããæ£ããèšéçµæžåŠã®ç¥èãããŒã¿åæã®ãªãã©ã·ãŒãäŒåŸããªããã°ãªããŸããðŠ
çŸå®ã®çµæžããŒã¿ããçè«ã¢ãã«ãšåœãŠã¯ã
æ£ããèšéææ³ã«ãã£ãŠå®èšŒåæã§ããåãéžæããã
ãã£ãšå°æ¥ã©ããã§æŽ»èºã§ãã人財ã«ãªããå¯èœæ§ãé«ããããšã«ç¹ãããšæããŸã
å®éã®çµæžååãæ¿æ²»ãšçµã³ã€ããªãã
å¿çšã§ããèœåããªããã°
ç¥èãæã€æ矩ãå°ãããªã£ãŠããŸããŸãðŠ
äœäºãã¢ãŠããããåæã®ã€ã³ãããã
倧äºã§ãããšãnoteã§æ¯æ¥çºä¿¡ããŠããŸãã
ããã¯ãã©ã®ãããªå
容ã§
ãã£ãŠãåœãŠã¯ãŸããŸãð
å
è¡ç 究ã®è«æãäžæŠã«èªãã§ã
èšæ¶ã«æ®ã£ãŠããªãã£ãã
倧åãªèŠ³ç¹ãå¿ããŠããŸã£ãŠãããããã
åŠç¿ã®é²æã¯æ»ã£ãŠããŸããšæããŸã
ã ããããããã®ãnoteãããã«æŽ»çšããŠ
èªåã®ç¥èã1ïŒ
ã§ããå®çãã
誰ã«ã§ããããããã解説ãã¢ãŠããããã§ããããã«åªããŠãããããšæããŸã
ç§ãããããã¢ãŠãããããã
èšéçµæžåŠã«ãããŠæéèŠãªããŒãã§ãã
æç³»ååæã®ã¢ãã«çè«è§£èª¬ã
ã©ããæåŸãŸã§ããæèªãã ããð
æ¬æçš¿äœæã«ãããåèæç®ã¯ä»¥äžã®éãã§ã
ãªããèšéçµæžåŠãåŠã¶ã®ãïŒïŒ
èšéçµæžåŠãæç³»å解ææ³ããçè«ãªãèšæž¬ããšããŠéãããšããããã¹ã¿ãŒãããããšã§ããç¥ãããŠããã®ã§ã
1930幎ã«åµç«ãããèšéçµæžåŠäŒã®èŠçŽç¬¬1æ¡ã§ã¯ãèšéçµæžåŠã¯ãçè«çæ°éçã¢ãããŒããšçµéšæ°éçã¢ãããŒãã®çµ±äžããšå®çŸ©ãããŠããŸããð
ãŸããRã»ããªãã·ã¥ã«ããããšã³ãã¡ããªã«ãåµåã®èŸã§ã¯ããçµ±èšåŠãçµæžåŠãæ°åŠã®äžè
ã®çµ±åããšå®çŸ©ãããŠããã®ã§ãð
ãã®ãããªå®çŸ©ã«ãããŠã¯ãåœæã®ããŒããŒãæ¯æ°äºæž¬ã«ä»£è¡šãããæç³»å解ææ³ãžã®æ¹å€ã匷ãæèãããŠãããšãããŠããŸã
ããªãã¡ãããã29幎ã®å€§ææ
ã®äºæž¬ã«å€±æããã®ã¯ïŒçµæžçè«ãç¡èŠãã æç³»åããŒã¿ã®åœ¢åŒçãªè§£æã®ã¿ã«çµå§ããããã§ãã£ããšããããšã§ã
ä»åŸã¯ãããããçè«ãªãèšæž¬ãã®ç«å Žãéãããçè«ã«åºã¥ãèšæž¬ããéèŠããŠãããªããã°ãªããªãããšããèŠè§£ã®éèŠæ§ãå¢ããŠããŸã
ãã®ãããªæŽå²ãçµãŠãèšéçµæžåŠã¯ã¹ã¿ãŒãããã£ãã®ã§ãã
ãããŠãäœãããã¯ãçµæžå€æ°ã¯
ãã®å€ããäºãã«åœ±é¿ãåãŒãåãçžäºäŸåã®é¢ä¿ã«ããããŸãéå»ã®å€åã®åœ±é¿ãæç¶ãããšããåŸåãæã¡ãŸã
ãããã®ååãåæããããå°æ¥ãäºæž¬ãããã§ããããã«ãªãããã«ã¯ãèšéçµæžåŠãã²ããŠã¯ãæç³»ååæãã«å¯Ÿããçè«ãæ£ããå®èšŒææ³ãžã®ç解ãå¿
èŠäžå¯æ¬ ãšãªããŸã
ãèšéçµæžåŠãã·ãªãŒãºã®æçš¿ã§ã¯ããããããã¯ãæç³»åå€æ°ã®å®èšŒåæã«å¿
èŠãªèšéçè«ãšææ³ãç¿åŸããããšãç®çãšããŸã
ããããç§ãã¢ãŠãããããã
æç³»åãã¯ãçµæžåæã«é¢ããå
容ã«ã€ããŠ
ã©ããæåŸãŸã§ãæèªãã ãããð
ãªããååã®èšäºã¯ãã¡ãã«ãªããŸãã®ã§ããã²ã埩ç¿ãã«ã掻çšãã ããã
æç³»ååæã®èª¬æã«ãããåºç€æŠå¿µð
æç³»åããŒã¿(time series data)ãšã¯ãæéã®çµéãšãšãã«èŠ³å¯ãããããŒã¿ãæããŸã
ããŒã¿ã®èŠ³æž¬é »åºŠã«ãã£ãŠãæ¥æ¬¡ããŒã¿ãæ次ããŒã¿ã幎次ããŒã¿ãªã©ããããŸã
ãããã£ãŠãæ¬ã·ãªãŒãºã®èª¬æã§ã¯ããã®ç¹ã匷調ããããã«ãå€æ°ã®æ·»ãåãå°æåã®tã§ãæšæ¬ã®å€§ããã倧æåã®Tã§è¡šèšããŸã
ãŸããæétã«å¯Ÿå¿ãã確çå€æ°ã®éå{yt}ãã確çéçš(stochastic process)ãšåŒã¶ããšã«ããŸã
$$
Stochastic Process\\ \\{ [y_t]}^â_{-â} = {[ âŠ,y_{-1},y_0,y_1,âŠ]}âŠ(1)
$$
ã€ãŸããæç³»åããŒã¿ãšã¯ç¢ºçéçšã«ãããŠ
第1æãã第Tæã«å¯Ÿå¿ãã確çå€æ°ã«ã€ããŠãå®çŸå€ãåŸããããã®ãšèããããŸã
æç³»ååæã«ãããŠéèŠãªåœ¹å²ãæ
ãã®ã¯ã次ã«èª¬æããå®åžžæ§(stationarity )ã®æŠå¿µã«ãªããŸã
ãªãã確çéçš{yt}ã«ã€ããŠã次ã®3ã€ã®æ§è³ªãæºãããããšãã{yt}ã¯å®åžžç(stationary)ã§ãããšè§£éããŸã
â»ããå³å¯ã«ã¯ã匱å®åžžæ§(weakly stationaly)ããŸãã¯å
±ååæ£æ€å®(covariance stationary) ã§ãããšèšãããŠããŸã
â å¹³å{yt}ãããã¹ãŠã®ïœã«å¯ŸããŠçããããš
â¡åæ£ïŒ¶{yt}ãããã¹ãŠã®ïœã«å¯ŸããŠçããããš
â¢èªå·±å
±åæ£Cov(yt,yt-1)ããæç¹ã®å·®ã§ããïœ(>0)ã®ã¿ã«äŸåããããš
ãã®3ã€ãæããããŸãã®ã§ããç解ãã ããã
$$
Stationary Conditions\\
Mean: E(y_t) =Ό \forall t \\
Variance:V(y_t) =Ï^2 \forall t \\ \\
Autocovariance:Cov(y_t,y_{t-1})\\
Only depending on time differential:s
$$
æç³»ååæã§ã¯ã2ã€ã®å€æ°y_tãšy_{t-s}ã®å ±åæ£ã以äžã®ããã«è¡šèšããèªå·±å ±åæ£(autocvariance)ãšåŒã³ãŸã
$$
γ(s) =Cov(y_t,y_{t-1}) , s>0 âŠ(2)
$$
ãŸããy_tãšy_{t-s}ã®èªå·±çžé¢ã以äžã®ããã«è¡šèšããããšã§ãèªå·±çžé¢é¢æ°(autocorrelation function:ACF)ãšåŒã¶ããšã«ããŸã
$$
Ï=Corr(y_t, y_{t-s}) \\ \\=\frac{Cov(y_t,y_{t-s})}{\sqrt{\smash[b]{V{(y_t)}V{(y_{t-s})}}} }=\frac{Cov(y_t,y_{t-s})}{V(y_t) }âŠ(3)
$$
ãŸããããã¯ã³ã¬ãã°ã©ã ãšåŒã°ããŠããããšãããã®ã§èŠããŠãããŠãã ããã
ãããŠãå®åžžç¢ºçéçšã¯ãèªå·±çžé¢é¢æ°ãŸãã¯èªå·±å
±åæ£é¢æ°ã«ãã£ãŠãã®ç¹åŸŽã瀺ãããã®ã§ã
ãã¯ã€ããã€ãºã®ååš
å®åžžç確çéçšã®ãªãã§ãæãç°¡åãªãã®ã¯ããã¯ã€ããã€ãº(white noise)ãšåŒã°ããŠããŸã
ããŸã誀差é
utããã¯ã€ããã€ãºãšãããšãããã¯ä»¥äžã®3ã€ã®æ§è³ªãæºããããšã«ãªããŸã
$$
u_t â¡White Noise \\ E(u_t)=0 \\V(u_t)=0\\Cov(u_t u_s)=0, tâ s
$$
ã€ãŸãããã¯ã€ããã€ãºã®ç¹åŸŽã¯
æåŸ
å€ïŒå¹³åïŒã0ã§ããããš
ãã€ãåæ£ãäžå®ã§ããããããŠ
èªå·±çžé¢ããªã確çå€æ°ã§ãããã®ã§ã
ãã ãããã¯ã€ããã€ãºéçšã§ã¯ãç¬ç«æ§ãä»®å®ãããŠããªãããšã«çæããå¿
èŠããããšæããŸã
ããã«å¯ŸããŠã確çå€æ°ããäºãã«ç¬ç«ã§åäžã®ååžã«åŸãããšããi.i.d.(independent and identically distributed) ãšè¡šçŸãããŸã
$$
Even if , E(u_t) =0 ,V(u_t)=Ï^2\\ \\u_t ïœi.i.d.(0,Ï^2)
$$
ããæåŸ å€ããŒãããã€åæ£ãäžå®ã§ã·ã°ãäºä¹(Ï^2)ã®å Žåã誀差é utãäºãã«ç¬ç«ã§åäžã®ååžã«åŸãããšãäžèšã®ããã«è¡šãã®ã§ãð
æ¬æ¥ã®è§£èª¬ã¯ä»¥äžãšããŸã
ãã¯ãçµæžåŠããããããããããã«
èšéçµæžåŠãªãã³ã«æç³»ååæã®ææ³ãäŒåŸããããšãå¿
èŠã§ãããšæããŸã
ä»åŸãšãäœåãå®ãããé¡ãããããŸãð
ä»é²ïŒç§ã®åè«ç 究ããŒãã«ã€ããŠð
ç§ã¯ãçºæ¿ä»å
¥ã®å®èšŒåæããããŒãã«
åæ¥è«æãå·çããããšèããŠããŸãð
æ¥æ¬çµæžãèãããšãã«ãçºæ¿ã¬ãŒãã«ãã£ãŠ
貿æååŒãçµåžžåæ¯ãå€åããã
æ ªã蚌åžãåµæš©ãšãã£ãéèè³ç£ã®åççã
å€åããããšæ¥æ¬çµæžãšçºæ¿ã¬ãŒããšã¯
åã£ãŠãåããªãçžãããã®ã§ãð
ïŒåðŽã ãã«ïœ¥ïœ¥ïœ¥ïŒ
çµæžã·ã§ãã¯ã«ãã£ãŠ
çºæ¿ã¬ãŒããå€åãããš
ãã®åœ±é¿ã¯ç§ãã¡ã®ç掻ã«å€§ãã圱é¿ããŸã
ã ãããããçºæ¿ã¬ãŒãã®å®å®æ§ã
æ
ä¿ãããããªçºæ¿ä»å
¥ã¯ãã¯ãçµæžæ¿çã«
ãããŠãéåžžã«éèŠãªæ矩ãæã£ãŠãããš
æšæž¬ããŠããŸã
決ããŠåŠéšçã楜ããŠå·çã§ããç°¡åãªããŒããéžæããŠããããã§ã¯ç¡ããšä¿¡ããŠããŸã
ãã ããã®åæ¥è«æãããåãããšã
ç§ã®åŠçç掻ã®é倧æãšãªãããšã¯äºå®ãªã®ã§
æåŸãŸã§ã³ãã³ããšåãçµãã§åããŸãð¥
æ¬æ¥ã®è§£èª¬ã¯ä»¥äžãšããŸã
ä»åŸãçµæžåŠçè«éãªãã³ã«
瀟äŒèª²é¡ã«å¯ŸããçµæžåŠçèŠç¹ã«ãã説æãªã©
ææ矩ãªå
容ãçºä¿¡ã§ããããã«åªããŠãŸãããŸãã®ã§ãä»åŸãšãå®ãããé¡ãããŸãð¥º
ãããããã¬ãžã³ã®ã玹ä»ð
ãã¡ãã®ãã¬ãžã³ã«ãŠ
åæ¥è«æå·çãžã®è»è·¡
ãšãã»ã³ã·ã£ã«çµæžåŠçè«éããªãã³ã«
åœéçµæžåŠðã®åºç€çè«ããŸãšããŠããŸã
ä»åŸãããã«ã³ã³ãã³ããæ¡å
ã§ããããã«åªããŠåããŸãã®ã§ãä»åŸãšãäœåãããããé¡ãç³ãäžããŸãð
ãŸãããã¡ãã«ïŒïŒåãšããŠã®ç§ã®å°±è·æŽ»åäœéšèšããŸãšãããã¬ãžã³ãã玹ä»ãããŠããã ããŸãð
æ§ã
ãªèŠ³ç¹ããå°±è·æŽ»åã«ã€ããŠèå¯ããŠããŸãã®ã§ããäžèªããã ããŸããšå¹žãã§ã
æ¹ããŠãå°±è·æŽ»åã¯
æ¬åœã«ããçžãã ãšæããŸããð
ã ããããããçžã倧åã«
ãããŠãéžãã éãæ£è§£ã«ã§ãããã
ãããããåªåããŠãããããªãšæããŸãð¥
æåŸãŸã§ãæèªããã ãèª ã«æé£ãããããŸããïŒ
ãããŸã§ãç§ã®èŠè§£ãæã£ãããšã
ãŸãšããããŠããã ããŠãŸãã
ãã®ç¹ã«é¢ããŸããŠããäºæ¿ãã ããð
ãã®æçš¿ãã¿ãŠãã ãã£ãæ¹ã
ã»ãã®å°ããªäºã§ãåŠã³ããã£ãïŒ
èãæ¹ã®åŒãåºããå¢ããïŒ
èªæžããåŠã¹ãããšãå€ãïŒ
ãªã©ãªã©ããã©ã¹ã®åç©«ããã£ãã®ã§ããã°
倧å€å¬ããæããŸãããæçš¿äœæã®å¥å©ã«å°œããŸãïŒïŒ
ãæ°è»œã«ã³ã¡ã³ããããããã¹ããð
ãããŠããå·®ãæ¯ããªããã°
ãã©ããŒïŒã·ã§ã¢ããé¡ããããã§ãð
ä»åŸãšãäœåãããããé¡ãããããŸãïŒ
ãã®èšäºãæ°ã«å ¥ã£ãããµããŒããããŠã¿ãŸãããïŒ