Asymptotic normality of the Nadaraya-Watson estimator for non-stationary functional data and applications to telecommunications.

Size: px
Start display at page:

Download "Asymptotic normality of the Nadaraya-Watson estimator for non-stationary functional data and applications to telecommunications."

Transcription

1 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary fuctioal data ad applicatios to telecommuicatios. L. ASPIROT, K. BERTIN, G. PERERA Departameto de Estadística, CIMFAV, Uiversidad de Valparaíso, Chile Istituto de Matemática y Estadística Rafael Laguardia, Facultad de Igeiería, Uruguay. July 6, 2007 Abstract We study a o-parametric regressio model where the explaatory variable is a ostatioary depedet fuctioal data ad the respose variable is scalar. Supposig that the explaatory variable is a o-statioary mixture of statioary processes ad geeral coditios of depedece of the observatios implied i particular by weakly depedece, we obtai the asymptotic ormality of the Nadaraya-Watso estimator. Uder some additioal regularity assumptios o the regressio fuctio, it ca be obtaied asymptotic cofidece itervals for the regressio fuctio. We apply this result to the estimatio of the quality of service o the Iteret where the cross traffic is o-statioary. Keywords: No-parametric regressio, fuctioal data, asymptotic ormality, o-statioarity, depedece, quality of service. AMS subject classificatio: 62G05, 62G08, 62G20. Itroductio We study the problem of estimatig a o-parametric regressio fuctio φ whe the explaatory variable is fuctioal ad may be o-statioary ad depedet. We observe {X, Y } = {X i, Y i : i = 0,..., } such that Y i = φx i + ε i, i = 0,...,, where φ is a fuctio φ : D R, X i D, Y i R, the ε i s are cetered ad idepedet of the X i s ad D is a semi-ormed liear space with semi-orm. Istituto de Matemática y Estadística Rafael Laguardia, IMERL, Facultad de Igeiería, Julio Herrera y Reissig 565, Motevideo 200, Tel-fax: , s: Departameto de Estadística, CIMFAV, Uiversidad de Valparaíso, Gra Bretaña 09, 4to piso, Playa Acha, Valparaíso, Chile, Tel: ,

2 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data 2 Our aim is to fid the asymptotic distributio of the estimator i= φ Y x Xi ik h x = 2 i= K x Xi h for x D where the kerel K is a positive fuctio ad the sequece h > 0 is the badwidth. We use the covetio 0/0 = 0. The estimator 2 is a geeralizatio of the Nadaraya-Watso estimator cf. []. The asymptotic ormality of the Nadaraya-Watso estimator is proved i [2] whe the observatios X i, Y i are idepedet, ad for depedet observatios, i [3]. Ferraty, Goia ad Vieu [4], [5] proposed the estimator 2 to estimate the regressio fuctio whe the explaatory variable is fuctioal. For weakly depedet ad statioary observatios, they proved the complete covergece of 2 ad obtaied rates of covergece whe the fuctio φ satisfies some regularity coditios. Aspirot et al [6] have geeralized the result of complete covergece for the estimator 2 to a o-statioary case. Masry 2005 [7] has proved the asymptotic ormality of φ for statioary weakly depedet radom variables. Ferraty, Mas ad Vieu 2007 [8] cosider asymptotic ormality of the estimator φ i the case of idepedece from theoretical ad practical poit of view. Related work o desity estimatio with fuctioal data ca be foud i [9] ad [0] for mode estimatio ad i [] for asymptotic ormality of kerel estimators. Surveys o fuctioal data aalysis ca be foud i Ramsay ad Silverma 2005 [2] ad i Ferraty ad Vieu 2006 [3] for the o-parametric case. Our goal i this article is to geeralize the result of [7] to the case whe X is a ostatioary mixture of statioary processes. More precisely, we suppose that there exist two radom processes ξ = {ξ : N}, Z = {Z : N} ad a fuctio ϕ : D R D such that ξ is statioary with values i D, ξ is idepedet of Z, Z takes values i a fiite set {z,...,z m } of R ad X satisfies X i = ϕξ i, Z i, i = 0,...,. 3 We suppose geeral coditios of depedece cf. assumptios i Subsectio 3., implied i particular by weakly depedece of the observatios ad of the process ξ, but the process Z may be o-statioary ad o-weakly depedet. I Theorem, uder these coditios ad a set of techical assumptios, we obtai the asymptotic ormality of the estimator 2. This result is based o cetral limit theorems for fields idexed i N or Z d. I Propositio ad 2, we prove results of cetral limit theorems for triagular arrays of a R m -valued cetered statioary radom field idexed i N or Z d. For these fields the existece of a cetral limit theorem depeds strogly o the geometry of the subset of Z d where the observatios are idexed. We apply these results to our model, i which we have a mixture of statioary fields X = ϕξ, Z where Z is o-statioary but X = ϕξ, Z coditioed to Z is statioary. We cosider here our observatios idexed i the level sets of Z. Some assumptios about the field Z should be made i order to esure properties of the level sets ad to have a cetral limit theorem i this case. I theorem 2, supposig additioally some regularity assumptios o the regressio fuctio φ ad o the badwidth h, we obtai the asymptotic ormality of φ x φx for x D, which allows us to costruct asymptotic cofidece iterval for φx.

3 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data 3 The o-statioary assumptio for the data appears aturally whe modellig the Iteret traffic cf. [4], [5], [6]. I this paper, we are iterested i estimatig the quality of service for voice ad video applicatios. We call quality of service parameters some etwork characteristics like delay, packet loss, etc.. The iterest i voice or video applicatios is based i the fact that these etwork traffic the same as may real time applicatios have high quality of service requiremets very low delay ad losses, etc.. I Sectio 4, we use the model 3 to estimate quality of service parameters for traffic over the Iteret. We assume that these parameters are a fuctio of the Iteret traffic, represeted by X = ϕξ, Z. I the literature, it is usual to assume that the traffic is piecewise statioary, i this model Z is a o-statioary radom process that selects betwee differet statioary behaviors. The process Z idicates for example if the etwork is very much loaded or ot ad it could describe seasoal or periodic behaviors. The process ξ is related with obtaied measuremets variatios. We give simulatios of the estimator 2 ad cofidece iterval, for simulated ad real traffic o the Iteret. The orgaizatio of the paper is as follows. I Sectio 2, we give some prelimiaries o the otios of asymptotically measurable sets ad fields ad state a cetral limit theorem for triagular arrays of R m -valued cetered statioary radom fields. I Sectio 3, we give our result o the asymptotic ormality of the estimator φ ad we make the assumptios uder which we obtai this result. Sectio 4 describes the applicatio of our result to the problem of estimatig the quality of service o the Iteret. Sectio 5 is devoted to the proofs. 2 Prelimiaries I this sectio we preset some prelimiary results. I subsectio 2. we defie the otio of asymptotically measurable subsets that allows, i subsectio 2.2, to give a cetral limit theorem for triagular arrays of R m -valued cetered statioary radom fields. I subsectio 2.3 we preset the otio of asymptotically measurable field that allows to state a cetral limit theorem for a field described by equatio 3. I this sectio, we cosider both the case whe the field is idexed i N ad i Z d ad i the followig, L stads for N or Z d. Whe we have L = N, we deote l = ad for a set A N, A = A {0,,..., } ad A c = A {0,,..., }. Whe we have L = Z d, we deote l = 2 + d ad for a set A Z d, A = A {,...,} d ad A c = A c {,...,} d. 2. Notio of asymptotically measurable subsets I what follows carda idicates the cardial of subset A. Defiitio. A subset A L is said to be a asymptotically measurable subset if, for all k L, the followig limit exists Fk, A = lim card{a c A k}. l The fuctio F., A is called the border fuctio of the subset A. Defiitio 2. The subset family { A i : i =,...,m } i L is a asymptotically measurable family if, for all i {,...,m}, the subsets A i are asymptotically measurable ad, for all k L ad

4 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data 4 i, j {,...,m}, the followig limit exists F k, A i, A j = lim card{a i A j k}. l 2.2 Cetral limit theorem for triagular arrays of R m -valued cetered statioary radom field I the followig, the otatio = w meas covergece i law ad N d 0, Σ represets a zero mea ormal distributio i dimesio d with covariace matrix Σ. Moreover, for X = Xk k L a R m -valued cetered statioary radom field, we deote the compoets of X by X = X,,...,X m, ad X k = X, k,...,xm, k for k L. Defiitio 3. We defie the class BL as the class of R m - valued cetered statioary radom fields X = X k k L that satisfies the followig coditios: H For all i, j {,...,m} ad N, we have { } E X i, 0 Xj, k <. k L H 2 Let X,J = X,,J,...,X m,,j be the trucatio by J of X, defied for k L ad J > 0 by [ ] X,J k = Xk { Xk J} E Xk { Xk J}, where represets the euclidia orm o R m. i There exists a sequece γk 0 such that k L γk < ad such that for all k L, N, i, j {,...,m} ad J > 0 we have E { X i,,j 0 X j,,j k } γk. ii There exists a sequece bj such that lim J bj = 0 ad for all set B L, i {,...,m}, N ad J > 0 where E [ S B, X i, S B, X i,,j ] 2 bjcardb, l S B, X i, = l /2 X i, k. k B H 3 There exists a sequece of reals umbers CJ such that for all B L, N, J > 0 ad i {,...,m} we have [ E S B, X i,,j ] 4 cardb 2 CJ. l

5 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data 5 H 4 There exist a fuctio h : R + R +, such that lim x + hx = 0 ad a fuctio g : R + R m R +, with gj, t < for all fixed J > 0 ad sup t R m gj, t = g J <, such that [ ] [ ] [ ] E e isb C, t,x,j E e isb, t,x,j E e isc, t,x,j gj, thdb, C, for all disjoit sets B, C L, all N, J > 0 ad t R m. Here, represets the scalar product o R m. H 5 There exist sequeces γ J i, j, k ad γi, j, k, i, j {,...,m}, k L such that for all J > 0 we have { } lim E X i,,j 0 X j,,j k = γ J i, j, k, ad for i, j {,...,m} ad k L. We have the two followig propositios proved i Sectio 5. lim J γj i, j, k = γi, j, k 4 Propositio. If X = X k k N belogs to BN, the for ay asymptotically measurable family { A i : i =,...,m } i N, we have as teds to where for i, j {,...,m} S A, X,,...,S A m, X m, w = N m 0, Σ, Σi, j = γi, j,0fi, j,0 + k {γi, j, kfi, j, k + γj, i, kfj, i, k} ad Fi, j, k = lim card{a i A j k}. Propositio 2. If X = X k k Z d belogs to BZd, the for ay asymptotically measurable family { A i : i =,...,m } i Z d, we have as teds to S A, X,,...,S A m, X m, w = N m 0, Σ, where for i, j {,...,m} ad Σi, j = k L γi, j, kfi, j, k card{a i A j k} Fi, j, k = lim 2 + d.

6 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data Notio of asymptotically measurable field I what follows a.s. meas almost sure covergece as teds to ad A is the idicator fuctio of a set A. I the case L = N, we set D = {0,..., } ad if L = Z d, we set D = {,...,} d. Defiitio 4. The R m -valued radom field Z = Z L is a asymptotically measurable field i L if there exists a radom probability measure R 0 i B, where B is the Borel σ-algebra i R m, such that a.s. R 0 B l m D {Zm B} ad for all k L {0} there exists a radom measure R k i B 2, where B 2 is the Borel σ-algebra i R 2m, such that for all B, C B l m D {Zm B} {Zm k C} a.s. R k B C. If the limit measure is o-radom Z is called a regular field i L. The followig propositio cf. Proof i [7] relates asymptotically measurable ad regular fields with asymptotically measurable subsets. Propositio 3. If Z = Z L is a asymptotically measurable field, B, B 2,...,B m are disjoit subsets of B ad A i = { L : Z B i } for i =,...,m, the, coditioally to Z the family {A,...,A m } is a asymptotically measurable family with Fk, A i, A j = R k B i, B j ad F0, A i, A j = R 0 B i δ ij, where δ i j is the Kroecker delta. Remarks: I the ext sectios we will assume further hypotheses o Z for the field described i 3 i order to prove the results, but the previous propositio is the tool for coditioig o level sets of the field Z ad applyig the results for statioary fields to the o-statioary case 3. 3 Results I the followig, we cosider a fixed x D ad we will study the asymptotic ormality of φ x give by 2. Proofs of the results are i Sectio 5. The estimator φ ca be also writte φ x = g x f x, where g x = f x = Y i K X i, ψh K X i 5 ψh

7 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data 7 with u x K u = K, u D. 6 ad ψh is give i assumptio A. We recall that we use the covetio 0/0 = 0. h I Subsectio 3., we give the assumptios uder which the asymptotic ormality of φ x is obtaied ad i Subsectio 3.2 we give our results. 3. Assumptios We make the followig assumptios o the distributio of X, Y. A There exist positive fuctios ψ, ψ,...,ψ m defied o R + D, fuctios c,...,c k defied o D ad a subset of {,...,m} such that for all h > 0 P [ ϕξ, z k x h] = c k xψ k h, x, ψ k h, x where lim h 0 ψh, x = if k, ad lim ψ k h, x h 0 ψh, x = 0 if k C. I the followig, to simplify the otatio, we replace ψh, x by ψh, but the quatity ψh still depeds of x. A 2 The fuctios u ψ k u, x are differetiable o R +, with differetial fuctio deoted ψ k u, x ad satisfy lim h 0 h ψ k h, x where the d k s are fuctios defied o D. 0 Kuψ k uh, xdu = d kx, A 3 The process Z is regular ad satisfies for k {,...,m} as teds to that ψh {Zi =z k } P PZ i = z k = 0, 7 where P = meas covergece i probability. For k {,...,m}, deote by p k the limit p k = lim PZ i = z k. Let the R 2m -value radom process X = X,,..., X 2m, defied for i N as follows: for l {,...,m} X l, i = ψh K ϕξ i, z l φϕξ i, z l + ε i ψh E [K ϕξ i, z l φϕξ i, z l ],

8 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data 8 for l {m +,...,2m} X l, i = where K is defied by 6. ψh K ϕξ i, z l Moreover we assume the followig hypotheses. A 4 The R 2m -value radom process X belogs to BN. A 5 The fuctio φ is cotiuous. ψh E [K ϕξ i, z l ], A 6 The estimator f x defied by 5 coverges i probability to fx > 0 where the fuctio f is defied by fu = k p k d k uc k u, u D. We suppose the followig hypothesis o the estimator φ. A 7 The fuctio K is a positive fuctio with support [0, ]. A 8 The badwidth h satisfies ad Remarks o the assumptios: lim h = 0 lim ψh =. The coditios A ad A 2 mea that, coditioally o Z, the distace betwee the observatio X i ad the estimatio poit x has a desity. The quatity ψ k h, x plays a role aalogous to h d for multidimesioal estimatio. The hypothesis A 3 meas that the variable Z does ot behave too irregularly. It is assumed that the process has some kid of statioarity i mea. However this is ot a strog assumptio, ad for example A 3 could be satisfied by periodic or semi-periodic radom variables. The relatio 7 is satisfied for example if the variables Zi =z k satisfy a cetral limit theorem for each k =,...,m. The hypothesis A 4 ca be replaced by various sets of hypothesis doig a trade-off betwee coditios o the momets of Y ad coditios of mixig. Perera 997 [8] describes several sets of coditios that imply the previous hypotheses. I the hypothesis A 6, fx > 0 meas roughly that there is a positive probability of fidig observatios X i ear i the sese of the semi-orm o D the poit x where we wat to compute the estimator. The fuctio f play the role of the desity of the whole process X = ϕξ, Z. Moreover the covergece i probability ca be obtaied uder the coditios A y A 2, coditios o the momets of Y i ad coditios of mixig cf. Proof i [6]. The hypothesis A 5, A 7 ad A 8 are classical for obtaiig asymptotic ormality.

9 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data Asymptotic ormality of φ x The asymptotic ormality of φ x is obtaied i three stages: asymptotic ormality of the field X i Propositio 4, of g x Eg x, f x Ef x i Propositio 5 ad the of φ x i Theorem. Theorem 2 gives the asymptotic ormality of φ x φx which allows to costruct asymptotic cofidece iterval for φx. Propositio 4. For l {,...,m}, deote A l = {k N, Z k = z l } ad A l+m = A l. If Z is asymptotically measurable, the uder Assumptio A 4, coditioally o Z, we have, as teds to S A, X,,..., S A 2m, X 2m, w = N 2m 0, Σ, where for i, j {,...,2m} Σi, j = γi, j,0fi, j,0 + k N {γi, j, kfi, j, k + γj, i, kfj, i, k}, with γi, j, k, k N, defied by 4 for the field X ad Fi, j, k = lim card{a i A j k}. Propositio 5. Uder Assumptios A 3 ad A 4, we have, as teds to, that w ψh g x Eg x, f x Ef x = N 2 0, A, where with a x = A = Σi, j, a 2 x = i,j= a x a 2 x a 2 x a 3 x 2 i= j=m+ Σi, j, a 3 x = 2 i,j=m+ Σi, j. Theorem. Uder Assumptios A A 8, we have as teds to ψh φ x Eg x w = N 0, σ 2 x, Ef x where Theorem 2. We suppose that B the fuctio φ satisfies with β > 0 ad L a positive costat, σ 2 x = a x 2a 2 xφx + a 3 xφ 2 x f 2. 8 x φu φv L u v β,

10 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data 0 B 2 there exist M > 0 ad a fuctio α such that for each k {,...,m} ad h > 0, we have P [ ϕξ, z k x h] c k x ψ k h, x Mαh α > 0. B 3 the badwidth h satisfies ad lim αh ψh = 0 lim hβ ψh = 0 B 4 for each k {,...,m}, we have lim ψh PZ i = z k p k = 0. The uder Assumptios A A 8, B B 4, we have as teds to, where σ 2 x is defied by 8. ψh φ x φx w = N 0, σ 2 x, 4 Applicatio to ed-to-ed quality of service estimatio 4. Motivatio ad model Nowadays ew services are offered over the Iteret like voice ad video. For these ew applicatios the eed to measure the etwork performace has icreased. Multimedia applicatios have several requiremets i terms of delay, losses ad other quality of service parameters. These costraits are stroger tha the oes for usual data trasfer applicatios mail, etc.. Measuremets of the Iteret performace are ecessary for differet reasos, for example to advace i uderstadig the behavior of the Iteret or to verify the quality of service assured to the ew services. There are several measurig techiques, most of them with software implemetatios, that ca be classified i active sedig cotrolled traffic called probe packets ad passive measures geerally measures at the routers. The aim of these techiques is to detect differet characteristics of the etwork that ca be the topology or some performace parameters delay, losses, available badwidth, etc.. There are also several problems to measure the performace parameters, for example the route of the packets ca chage, the traffic bit rate is ot costat ad ormally is i bursts, the probe packets ca be filtered or altered by oe ISP Iteret Service Provider i the path, there is ot clock sychroizatio betwee routers ad ed equipmets, etc.. Normally the iteral routers i the path betwee two poits of iterest are ot uder the cotrol of oly oe user or oe ISP. Therefore, it is ot very useful to have measurig procedures that deped o the iformatio of the iteral routers. For this reaso ed-to-ed measures is oe of the most developed methodologies durig the last years cf. [9], [20],[2]. We cosider a sigle lik ad a voice or video traffic that we wat to moitor, that is we wat to kow the quality of service parameters whe this traffic goes betwee two poits over

11 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data the etwork. The lik is shared by this multimedia traffic ad may other traffics i the etwork that are ukow ad that is called cross traffic ad could be modelled as a stochastic process. The performace parameter for packets of a multimedia traffic could be delay, packets losses, delay variatio or all of them together ad we will cosider it as a radom variable Y. This performace parameter could be modeled as a fuctio of the cross traffic stochastic process, the video or voice stochastic process, the lik capacity ad the buffer size. We wat to estimate Y without sedig video or voice traffic durig log time periods. Active measuremets, that sed traffic, are useful if they do ot charge the etwork, so they cosist o small probe packets set durig short time periods. The lik capacity ad the buffer size are ot kow but it is assumed that they are costats durig the moitorig process. The multimedia traffic is also kow therefore we ca suppose that Y = φx where X is a characterizatio of the cross traffic stochastic process. We ca ot measure the cross traffic process but we ca obtai data closely related with this process. This data is described i the followig subsectio. The cross traffic process o the Iteret is a depedet o-statioary process c.f [4] or [5]. To take ito accout o-statioary behaviors we will suppose that X satisfies the relatio Measuremet procedure I what follows we describe the procedure to estimate the fuctio φ. We divide the experimet i two phases: first, we sed a burst of small probe packets pp of fixed size spaced by a fixed time. Immediately after the burst we sed durig a short time a video stream. We repeat the previous procedure durig some time lapse, sedig a ew burst ad a video probe after a time iterval measured from the previous ed of the video stream as it is show i the scheme i figure. pp video Figure : Measuremet procedure scheme With the probe packets burst we ifer the cross traffic of the lik. We measure at the output of the lik the iterarrival time betwee cosecutive probe packets. This time serie is strogly correlated with the cross traffic process that shares the lik with the probe traffic. We compute the empirical distributio fuctio of probe packets iterarrival times because this distributio is related with the queue behavior i the etwork. Usig this cross traffic estimatio ad measurig the delay Y over the video, we wat to estimate the fuctio φ. From each probe packet burst ad video sequece j we have a pair X j, Y j, where X j is the empirical distributio fuctio of iterarrival times ad Y j is the performace metric of iterest measured from the video stream j. Therefore, our estimatio problem has bee trasformed ito the problem of iferrig a fuctio φ : D R where D is the space of the probability distributio fuctios ad R is the real lie such that the X j, Y j satisfy.

12 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data 2 We did the estimatios both with simulated ad real traffic. Real data are obtaied with a software that do the active measures described before betwee a server ad its users. For simulated data, the variable Y is the oe-way delay which is the delay betwee two poits. For real data, we measure roud trip time RTT delay that is the delay i the whole trip betwee two poits i a path ad i the reverse path. For simulated traffic, we have 360 observatios X j, Y j. For each j {,...,359}, we calculate the estimator φ j defied by 2 usig the observatios X, Y,...,X j, Y j ad we represet o figures 2 ad 3 the values of Y j+ ad of its predictio φ j X j+. Moreover for each j {,...,359}, we calculate a cofidece iterval aroud φ j X j+ usig a block-bootstrap. The estimator φ j is calculated usig the kerel Kx = x 2 2 [,] x, the badwidth that miimizes the sum of the relative error i the estimatio before the time j ad the L orm. The figure 4 represets the relative error of the estimatio. The results are relatively accurate for sample of size larger tha 00. For real data, we have 50 observatios X j, Y j ad we realize the same type of estimatio. I figure 5, we represets for each j {,...,49}, the values of Y j+ ad of its predictio φ j X j+ ad a cofidece a cofidece bad aroud φ j X j+ usig a block-bootstrap. The figure 6 represets the relative error of the estimatio. The results are less accurate tha i the case of simulated data but the estimatios are calculated with samples of small size Delayms Sample umber Figure 2: Estimated delays for all simulated data

13 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data Predictio delay Measured delay Upper boud cofidece Lower boud cofidece Delayms Sample umber Figure 3: Estimated delays for simulated data observatios betwee 230 ad Relative error Sample umber Figure 4: Relative error for simulated data

14 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data Predictio delay Measured delay Upper boud cofidece Lower boud cofidece 200 Delayms Sample umber Figure 5: Estimated RTT for real data Relative error Sample umber Figure 6: Relative error for real data

15 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data 5 5 Proof of the propositios ad theorems 5. Proof of Propositios ad 2 This result is obtaied by meas of Bershtei method cf. [8], [22] ad it is a geeralizatio of the theorem 4.5 o triagular arrays for R-valued radom fields obtaied by Tablar 2006 [23], p 86 to R m -valued radom fields. To prove that the vectorial field S A, X,,..., S A m, X m, w = N m 0, Σ, it is sufficiet to prove that for all λ R m w λ, S = N0, λ t Σλ where S = S A, X,,..., S A m, X m, ad λ t = λ,...,λ m. We have that S, λ = = λ i S A i, X i, = i= l /2 k D i= i= l /2λ i λ i X i, k {k A i } Let us cosider a real valued field X λ where X i, k λ = λ ix i, k if k A i. The S, λ = S D, X λ. As the field X λ verifies the hypotheses i the work of [23] we ca apply the theorem for a real valued radom field i order to obtai the result of propositios ad Proof of Propositio 4 Sice Z is asymptotically measurable, coditioally to Z, the subset family A,...,A 2m is a asymptotically measurable family i N. The applyig Propositio to X which belogs to BN, we deduce the result. 5.3 Proof of Propositio 5 Before provig Propositio 5, here we prove the followig lema. Lemma. Let k {,...,m}. Uder the Assumptios A, A 2, A 3, A 5, A 7 ad A 8, we have the followig limit lim k A i X i, k ψh E [K ϕξ, z k ] = d k xc k x {k }, 2 for β > 0, lim ψh E [ ϕξ, z k x h β K ϕξ, z k ] d k xc k x, 3 that the estimator f x satisfies lim Ef x = fx,

16 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data 6 4 ad the estimator g x satisfies lim Eg x = φxfx. Proof. The results ad 2 of this lemma come from the fact that uder Assumptio A, the desity of ϕξ, z k x is the fuctio u c k xψ k u, x. The uder Assumptio A 7 ad [ ϕξ, z k x E h E [K ϕξ, z k ] = h c k x β K ϕξ, z k ] 0 = h c k x Kuψ k uh, xdu ad it is sufficiet to use Assumptios A 2 ad A 8 to coclude. Here we prove 3 ad 4. We have for i {,...,}, EK X i = E{EK X i Z i } = 0 Kuu β ψ k uh, xdu E{K ϕξ i, z k }PZ i = z k. k= As ϕξ i, z k i=,..., is idetically distributed, we obtai that, Ef x = ψh E K ϕξ, z k PZ i = z k. 9 k= Now applyig A 3 ad the assertio of this lemma i 9, we obtai 3. Similar calculus give that Eg x = ψh E φϕξ, z k K ϕξ, z k PZ i = z k. The we have where R = k= Eg x = φxef x + R, ψh E {φϕξ, z k φx} K ϕξ, z k PZ i = z k k= sup u: x u h φu φx Ef x ad we obtai 4 usig the cotiuity of fuctio φ Assumptio A 5 ad usig that lim Ef x = fx. Coditioally to Z, we have ψh g x Eg x Z, f x Ef x Z = S A j, X,j, j= 2 j=m+ S A j, X,j.

17 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data 7 Usig Propositio 4, we obtai that, coditioally to Z, S A j, X,j, 2 S A j, X,j = w N 2 0, A. j= j=m+ Sice Z is regular, the matrix A is ot radom ad the we have w ψh g x Eg x Z, f x Ef x Z = N 2 0, A. Now g x Eg x, f x Ef x = g x Eg x Z, f x Ef x Z We have Ef x Z Ef x = 2 j= + Eg x Z Eg x, Ef x Z Ef x. EK ξ, z j ψh carda j P[Z i = j] Uder Assumptio A 3 ad usig assertio of Lemma, we have that ψh Ef x Z Ef x. coverges i probability to 0 as teds to. Usig Assumptio A 3, ψh Eg x Z Eg x coverges i probability to 0 as teds to. Usig the lemma of Slutsky, we obtai the result of the propositio. 5.4 Proof of Theorem We have where ad φ x Eg x Ef x = Q x B xf x E f x, f x B x = Eg x Ef x φx Q x = g x Eg x φxf x Ef x. Usig 3 y 4 of Lemma, we deduce that lim B x = 0. The usig A 6, we have that f x coverges i probability to fx > 0 ad the it implies that B xf x E f x 0 f x coverges i probability to 0. Fially, usig A 6, 0 ad Propositio 5, applyig Slutsky Lemma, we deduce the result of the propositio.

18 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data Proof of Theorem 2 Theorem 2 is obtaied usig Theorem ad the followig lemma. Lemma 2. Uder Assumptios B B 4 ad Assumptios A, A 2, A 6, A 7 ad A 8, we have that Eg x lim ψh Ef x φx = 0. Proof. We have the followig decompositio Eg x Ef x φx = Ef x [Eg x fxφx] + φx Ef x [fx Ef x]. Here we study first the quatity Ef x fx. As i Lemma, we have Ef x = ψh E K ϕξ, z k PZ i = z k ad the Ef x fx = + k= k= ψh E K ϕξ, z k d k xc k x {k } PZ i = z k d k xc k x {k } PZ i = z k p k. k= Uder the Assumptios B 2, B 3 ad B 4 imply that ψh Ef x fx = 0. 2 lim Here we study the term Eg x fxφx. It satisfies Eg x fxφx φx Ef x fx + Eg x Ef xφx. We have Eg x Ef xφx = E [K ϕξ, z k φϕξ, z k φx] PZ i = z k ψh k= [ Lh β ψh E K ϕξ, z k ϕξ, z k x ] β h PZ i = z k k= L h β + o, 3 where the secod lie comes from Assumptio B ad the third is a cosequece of assertio 2 of Lemma. From 2, 3 ad Assumptio B 3, we deduce that ψh Eg x fxφx = 0. 4 lim Assertio of Lemma, 2 ad 4 applyig i imply the propositio.

19 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data 9 6 Ackowledgemet This work has bee supported by the ARTES group. Karie Berti has bee supported by Project DIPUV 33/05, Project FONDECYT 0684 ad the Laboratory ANESTOC ACT- 3. Laura Aspirot has bee partially supported by PDT Programa de Desarrollo Tecológico S/C/OP/46/. The authors would like to thak Yoaa Dooso for her cotributios to the simulatios programs. Refereces [] Nadaraya, E. A., 989, Noparametric estimatio of probability desities ad regressio curves, Mathematics ad its Applicatios Soviet Series, 20. Dordrecht: Kluwer Academic Publishers Group. [2] Roseblatt, M., 969, Coditioal probability desity ad regressio estimators. Multivariate Aalysis, II Proc. Secod Iterat. Sympos., Dayto, Ohio, 968, New York: Academic Press. [3] Robiso, P. M., 983, Noparametric estimators for time series. J. Time Ser. Aal., 4, [4] Ferraty, F., Goia, A. ad Vieu, P., 2002, Fuctioal oparametric model for time series: a fractal approach for dimesio reductio. Test,, [5] Ferraty, F. ad Vieu, P., 2004, Noparametric models for fuctioal data, with applicatio i regressio, time-series predictio ad curve discrimiatio. J. Noparametr. Stat., 6, 25. [6] Aspirot, L., Belzarea, P., Bazzao, B. ad Perera, G., 2005, Ed-To-Ed Quality of Service Predictio Based O Fuctioal Regressio. Proc. Third Iteratioal Workig Coferece o Performace Modellig ad Evaluatio of Heterogeeous Networks HET-NETs 2005, Ilkley, UK, July [7] Masry, E., 2005, Noparametric regressio estimatio for depedet fuctioal data: asymptotic ormality. Stochastic Process. Appl., 5, [8] Ferraty, F., Mas, A. ad Vieu, P., 2007, Noparametric regressio o fuctioal data: iferece ad pratical aspects. Australia ad New Zealad Joural of Statistics,. [9] Gasser, T., Hall, P. ad Presell, B., 998, Noparametric estimatio of the mode of a distributio of radom curves. J. R. Stat. Soc. Ser. B Stat. Methodol., 60, [0] Dabo-Niag, S., Ferraty, F., ad Vieu, P., 2004, Estimatio du mode das u espace vectoriel semi-ormé. C. R. Math. Acad. Sci. Paris, 339, [] Oliveira, P., 2005, Noparametric desity ad regressio estimatio fuctioal data. Pré- Publicaçoes do Departameto de Matemática, Uiversidade de Coimbra, [2] Ramsay, J. O. ad Silverma, B. W., 2005, Fuctioal data aalysis 2d ed, Spriger Series i Statistics. New York: Spriger.

20 Asymptotic ormality of the Nadaraya-Watso estimator for o-statioary data 20 [3] Ferraty, F. ad Vieu, P., 2006, Noparametric Fuctioal data aalysis: Theory ad Practice, Spriger Series i Statistics. New York: Spriger. [4] Karagiais, T., Molle, M., Faloutsos, M. ad Broido, A., 2004, A ostatioary Poisso view of Iteret traffic. Proc. of INFOCOM Twety-third Aual Joit Coferece of the IEEE Computer ad Commuicatios Societies, 3, [5] Zhag, Y., Paxso, V. ad Sheker, S., 2000, The Statioarity of Iteret Path Properties: Routig, Loss, ad Throughput. ACIRI Techical Report. [6] Zhag, Y. ad Duffield, N., 200, O the costacy of iteret path properties. IMW 0: Proceedigs of the st ACM SIGCOMM Workshop o Iteret Measuremet, [7] Perera, G., 2002, Irregular sets ad cetral limit theorems. Beroulli, 8, [8] Perera, G., 997, Geometry of Z d ad the cetral limit theorem for weakly depedet radom fields. J. Theoret. Probab., 0, [9] Bolot, J., 993, Ed-to-ed packet delay ad loss behavior i the iteret. SIGCOMM 93: Coferece proceedigs o Commuicatios architectures, protocols ad applicatios, [20] Jai, M. ad Dovrolis, C., 2002, Pathload: A measuremet tool for ed-to-ed available badwidth. Proc. of Passive ad Active Measuremets PAM Workshop, Mar [2] Corral, J., Texier, G. ad Toutai, L., 2003, Ed-to-ed Active Measuremet Architecture i IP Networks SATURNE. PAM 2003, A workshop o Passive ad Active Measuremets. [22] Perera, G., 997, Applicatios of cetral limit theorems over asymptotically measurable sets: regressio models. C. R. Acad. Sci. Paris Sér. I Math., 324, [23] Tablar, A., Estadística de procesos estocásticos y Aplicacioes a Modelos Epidémicos, PhD tesis, submitted to Departameto de Graduados de la Uiversidad Nacioal del Sur for evaluatio, december 2006.

Properties of MLE: consistency, asymptotic normality. Fisher information.

Properties of MLE: consistency, asymptotic normality. Fisher information. Lecture 3 Properties of MLE: cosistecy, asymptotic ormality. Fisher iformatio. I this sectio we will try to uderstad why MLEs are good. Let us recall two facts from probability that we be used ofte throughout

More information

I. Chi-squared Distributions

I. Chi-squared Distributions 1 M 358K Supplemet to Chapter 23: CHI-SQUARED DISTRIBUTIONS, T-DISTRIBUTIONS, AND DEGREES OF FREEDOM To uderstad t-distributios, we first eed to look at aother family of distributios, the chi-squared distributios.

More information

Output Analysis (2, Chapters 10 &11 Law)

Output Analysis (2, Chapters 10 &11 Law) B. Maddah ENMG 6 Simulatio 05/0/07 Output Aalysis (, Chapters 10 &11 Law) Comparig alterative system cofiguratio Sice the output of a simulatio is radom, the comparig differet systems via simulatio should

More information

3. Covariance and Correlation

3. Covariance and Correlation Virtual Laboratories > 3. Expected Value > 1 2 3 4 5 6 3. Covariace ad Correlatio Recall that by takig the expected value of various trasformatios of a radom variable, we ca measure may iterestig characteristics

More information

Chapter 7 Methods of Finding Estimators

Chapter 7 Methods of Finding Estimators Chapter 7 for BST 695: Special Topics i Statistical Theory. Kui Zhag, 011 Chapter 7 Methods of Fidig Estimators Sectio 7.1 Itroductio Defiitio 7.1.1 A poit estimator is ay fuctio W( X) W( X1, X,, X ) of

More information

7. Sample Covariance and Correlation

7. Sample Covariance and Correlation 1 of 8 7/16/2009 6:06 AM Virtual Laboratories > 6. Radom Samples > 1 2 3 4 5 6 7 7. Sample Covariace ad Correlatio The Bivariate Model Suppose agai that we have a basic radom experimet, ad that X ad Y

More information

Asymptotic Growth of Functions

Asymptotic Growth of Functions CMPS Itroductio to Aalysis of Algorithms Fall 3 Asymptotic Growth of Fuctios We itroduce several types of asymptotic otatio which are used to compare the performace ad efficiecy of algorithms As we ll

More information

An example of non-quenched convergence in the conditional central limit theorem for partial sums of a linear process

An example of non-quenched convergence in the conditional central limit theorem for partial sums of a linear process A example of o-queched covergece i the coditioal cetral limit theorem for partial sums of a liear process Dalibor Volý ad Michael Woodroofe Abstract A causal liear processes X,X 0,X is costructed for which

More information

ORDERS OF GROWTH KEITH CONRAD

ORDERS OF GROWTH KEITH CONRAD ORDERS OF GROWTH KEITH CONRAD Itroductio Gaiig a ituitive feel for the relative growth of fuctios is importat if you really wat to uderstad their behavior It also helps you better grasp topics i calculus

More information

NPTEL STRUCTURAL RELIABILITY

NPTEL STRUCTURAL RELIABILITY NPTEL Course O STRUCTURAL RELIABILITY Module # 0 Lecture 1 Course Format: Web Istructor: Dr. Aruasis Chakraborty Departmet of Civil Egieerig Idia Istitute of Techology Guwahati 1. Lecture 01: Basic Statistics

More information

Chapter 7 - Sampling Distributions. 1 Introduction. What is statistics? It consist of three major areas:

Chapter 7 - Sampling Distributions. 1 Introduction. What is statistics? It consist of three major areas: Chapter 7 - Samplig Distributios 1 Itroductio What is statistics? It cosist of three major areas: Data Collectio: samplig plas ad experimetal desigs Descriptive Statistics: umerical ad graphical summaries

More information

Case Study. Normal and t Distributions. Density Plot. Normal Distributions

Case Study. Normal and t Distributions. Density Plot. Normal Distributions Case Study Normal ad t Distributios Bret Halo ad Bret Larget Departmet of Statistics Uiversity of Wiscosi Madiso October 11 13, 2011 Case Study Body temperature varies withi idividuals over time (it ca

More information

In nite Sequences. Dr. Philippe B. Laval Kennesaw State University. October 9, 2008

In nite Sequences. Dr. Philippe B. Laval Kennesaw State University. October 9, 2008 I ite Sequeces Dr. Philippe B. Laval Keesaw State Uiversity October 9, 2008 Abstract This had out is a itroductio to i ite sequeces. mai de itios ad presets some elemetary results. It gives the I ite Sequeces

More information

Hypothesis Tests Applied to Means

Hypothesis Tests Applied to Means The Samplig Distributio of the Mea Hypothesis Tests Applied to Meas Recall that the samplig distributio of the mea is the distributio of sample meas that would be obtaied from a particular populatio (with

More information

Characterizing End-to-End Packet Delay and Loss in the Internet

Characterizing End-to-End Packet Delay and Loss in the Internet Characterizig Ed-to-Ed Packet Delay ad Loss i the Iteret Jea-Chrysostome Bolot Xiyu Sog Preseted by Swaroop Sigh Layout Itroductio Data Collectio Data Aalysis Strategy Aalysis of packet delay Aalysis of

More information

Maximum Likelihood Estimators.

Maximum Likelihood Estimators. Lecture 2 Maximum Likelihood Estimators. Matlab example. As a motivatio, let us look at oe Matlab example. Let us geerate a radom sample of size 00 from beta distributio Beta(5, 2). We will lear the defiitio

More information

Confidence Intervals for One Mean with Tolerance Probability

Confidence Intervals for One Mean with Tolerance Probability Chapter 421 Cofidece Itervals for Oe Mea with Tolerace Probability Itroductio This procedure calculates the sample size ecessary to achieve a specified distace from the mea to the cofidece limit(s) with

More information

A probabilistic proof of a binomial identity

A probabilistic proof of a binomial identity A probabilistic proof of a biomial idetity Joatho Peterso Abstract We give a elemetary probabilistic proof of a biomial idetity. The proof is obtaied by computig the probability of a certai evet i two

More information

Our aim is to show that under reasonable assumptions a given 2π-periodic function f can be represented as convergent series

Our aim is to show that under reasonable assumptions a given 2π-periodic function f can be represented as convergent series 8 Fourier Series Our aim is to show that uder reasoable assumptios a give -periodic fuctio f ca be represeted as coverget series f(x) = a + (a cos x + b si x). (8.) By defiitio, the covergece of the series

More information

1 Computing the Standard Deviation of Sample Means

1 Computing the Standard Deviation of Sample Means Computig the Stadard Deviatio of Sample Meas Quality cotrol charts are based o sample meas ot o idividual values withi a sample. A sample is a group of items, which are cosidered all together for our aalysis.

More information

Hypothesis testing. Null and alternative hypotheses

Hypothesis testing. Null and alternative hypotheses Hypothesis testig Aother importat use of samplig distributios is to test hypotheses about populatio parameters, e.g. mea, proportio, regressio coefficiets, etc. For example, it is possible to stipulate

More information

1 Correlation and Regression Analysis

1 Correlation and Regression Analysis 1 Correlatio ad Regressio Aalysis I this sectio we will be ivestigatig the relatioship betwee two cotiuous variable, such as height ad weight, the cocetratio of a ijected drug ad heart rate, or the cosumptio

More information

5: Introduction to Estimation

5: Introduction to Estimation 5: Itroductio to Estimatio Cotets Acroyms ad symbols... 1 Statistical iferece... Estimatig µ with cofidece... 3 Samplig distributio of the mea... 3 Cofidece Iterval for μ whe σ is kow before had... 4 Sample

More information

Modified Line Search Method for Global Optimization

Modified Line Search Method for Global Optimization Modified Lie Search Method for Global Optimizatio Cria Grosa ad Ajith Abraham Ceter of Excellece for Quatifiable Quality of Service Norwegia Uiversity of Sciece ad Techology Trodheim, Norway {cria, ajith}@q2s.tu.o

More information

0.7 0.6 0.2 0 0 96 96.5 97 97.5 98 98.5 99 99.5 100 100.5 96.5 97 97.5 98 98.5 99 99.5 100 100.5

0.7 0.6 0.2 0 0 96 96.5 97 97.5 98 98.5 99 99.5 100 100.5 96.5 97 97.5 98 98.5 99 99.5 100 100.5 Sectio 13 Kolmogorov-Smirov test. Suppose that we have a i.i.d. sample X 1,..., X with some ukow distributio P ad we would like to test the hypothesis that P is equal to a particular distributio P 0, i.e.

More information

Economics 140A Confidence Intervals and Hypothesis Testing

Economics 140A Confidence Intervals and Hypothesis Testing Ecoomics 140A Cofidece Itervals ad Hypothesis Testig Obtaiig a estimate of a parameter is ot the al purpose of statistical iferece because it is highly ulikely that the populatio value of a parameter is

More information

SAMPLE QUESTIONS FOR FINAL EXAM. (1) (2) (3) (4) Find the following using the definition of the Riemann integral: (2x + 1)dx

SAMPLE QUESTIONS FOR FINAL EXAM. (1) (2) (3) (4) Find the following using the definition of the Riemann integral: (2x + 1)dx SAMPLE QUESTIONS FOR FINAL EXAM REAL ANALYSIS I FALL 006 3 4 Fid the followig usig the defiitio of the Riema itegral: a 0 x + dx 3 Cosider the partitio P x 0 3, x 3 +, x 3 +,......, x 3 3 + 3 of the iterval

More information

if A S, then X \ A S, and if (A n ) n is a sequence of sets in S, then n A n S,

if A S, then X \ A S, and if (A n ) n is a sequence of sets in S, then n A n S, Lecture 5: Borel Sets Topologically, the Borel sets i a topological space are the σ-algebra geerated by the ope sets. Oe ca build up the Borel sets from the ope sets by iteratig the operatios of complemetatio

More information

MATH 361 Homework 9. Royden Royden Royden

MATH 361 Homework 9. Royden Royden Royden MATH 61 Homework 9 Royde..9 First, we show that for ay subset E of the real umbers, E c + y = E + y) c traslatig the complemet is equivalet to the complemet of the traslated set). Without loss of geerality,

More information

Module 4: Mathematical Induction

Module 4: Mathematical Induction Module 4: Mathematical Iductio Theme 1: Priciple of Mathematical Iductio Mathematical iductio is used to prove statemets about atural umbers. As studets may remember, we ca write such a statemet as a predicate

More information

Sequences and Series

Sequences and Series CHAPTER 9 Sequeces ad Series 9.. Covergece: Defiitio ad Examples Sequeces The purpose of this chapter is to itroduce a particular way of geeratig algorithms for fidig the values of fuctios defied by their

More information

Measurable Functions

Measurable Functions Measurable Fuctios Dug Le 1 1 Defiitio It is ecessary to determie the class of fuctios that will be cosidered for the Lebesgue itegratio. We wat to guaratee that the sets which arise whe workig with these

More information

Non-life insurance mathematics. Nils F. Haavardsson, University of Oslo and DNB Skadeforsikring

Non-life insurance mathematics. Nils F. Haavardsson, University of Oslo and DNB Skadeforsikring No-life isurace mathematics Nils F. Haavardsso, Uiversity of Oslo ad DNB Skadeforsikrig Mai issues so far Why does isurace work? How is risk premium defied ad why is it importat? How ca claim frequecy

More information

Section IV.5: Recurrence Relations from Algorithms

Section IV.5: Recurrence Relations from Algorithms Sectio IV.5: Recurrece Relatios from Algorithms Give a recursive algorithm with iput size, we wish to fid a Θ (best big O) estimate for its ru time T() either by obtaiig a explicit formula for T() or by

More information

Definition. Definition. 7-2 Estimating a Population Proportion. Definition. Definition

Definition. Definition. 7-2 Estimating a Population Proportion. Definition. Definition 7- stimatig a Populatio Proportio I this sectio we preset methods for usig a sample proportio to estimate the value of a populatio proportio. The sample proportio is the best poit estimate of the populatio

More information

FIBONACCI NUMBERS: AN APPLICATION OF LINEAR ALGEBRA. 1. Powers of a matrix

FIBONACCI NUMBERS: AN APPLICATION OF LINEAR ALGEBRA. 1. Powers of a matrix FIBONACCI NUMBERS: AN APPLICATION OF LINEAR ALGEBRA. Powers of a matrix We begi with a propositio which illustrates the usefuless of the diagoalizatio. Recall that a square matrix A is diogaalizable if

More information

8.5 Alternating infinite series

8.5 Alternating infinite series 65 8.5 Alteratig ifiite series I the previous two sectios we cosidered oly series with positive terms. I this sectio we cosider series with both positive ad egative terms which alterate: positive, egative,

More information

Chapter 6: Variance, the law of large numbers and the Monte-Carlo method

Chapter 6: Variance, the law of large numbers and the Monte-Carlo method Chapter 6: Variace, the law of large umbers ad the Mote-Carlo method Expected value, variace, ad Chebyshev iequality. If X is a radom variable recall that the expected value of X, E[X] is the average value

More information

Section 7-3 Estimating a Population. Requirements

Section 7-3 Estimating a Population. Requirements Sectio 7-3 Estimatig a Populatio Mea: σ Kow Key Cocept This sectio presets methods for usig sample data to fid a poit estimate ad cofidece iterval estimate of a populatio mea. A key requiremet i this sectio

More information

Overview of some probability distributions.

Overview of some probability distributions. Lecture Overview of some probability distributios. I this lecture we will review several commo distributios that will be used ofte throughtout the class. Each distributio is usually described by its probability

More information

LECTURE 13: Cross-validation

LECTURE 13: Cross-validation LECTURE 3: Cross-validatio Resampli methods Cross Validatio Bootstrap Bias ad variace estimatio with the Bootstrap Three-way data partitioi Itroductio to Patter Aalysis Ricardo Gutierrez-Osua Texas A&M

More information

Class Meeting # 16: The Fourier Transform on R n

Class Meeting # 16: The Fourier Transform on R n MATH 18.152 COUSE NOTES - CLASS MEETING # 16 18.152 Itroductio to PDEs, Fall 2011 Professor: Jared Speck Class Meetig # 16: The Fourier Trasform o 1. Itroductio to the Fourier Trasform Earlier i the course,

More information

Irreducible polynomials with consecutive zero coefficients

Irreducible polynomials with consecutive zero coefficients Irreducible polyomials with cosecutive zero coefficiets Theodoulos Garefalakis Departmet of Mathematics, Uiversity of Crete, 71409 Heraklio, Greece Abstract Let q be a prime power. We cosider the problem

More information

THE REGRESSION MODEL IN MATRIX FORM. For simple linear regression, meaning one predictor, the model is. for i = 1, 2, 3,, n

THE REGRESSION MODEL IN MATRIX FORM. For simple linear regression, meaning one predictor, the model is. for i = 1, 2, 3,, n We will cosider the liear regressio model i matrix form. For simple liear regressio, meaig oe predictor, the model is i = + x i + ε i for i =,,,, This model icludes the assumptio that the ε i s are a sample

More information

Stat 104 Lecture 16. Statistics 104 Lecture 16 (IPS 6.1) Confidence intervals - the general concept

Stat 104 Lecture 16. Statistics 104 Lecture 16 (IPS 6.1) Confidence intervals - the general concept Statistics 104 Lecture 16 (IPS 6.1) Outlie for today Cofidece itervals Cofidece itervals for a mea, µ (kow σ) Cofidece itervals for a proportio, p Margi of error ad sample size Review of mai topics for

More information

N04/5/MATHL/HP2/ENG/TZ0/XX MATHEMATICS HIGHER LEVEL PAPER 2. Thursday 4 November 2004 (morning) 3 hours INSTRUCTIONS TO CANDIDATES

N04/5/MATHL/HP2/ENG/TZ0/XX MATHEMATICS HIGHER LEVEL PAPER 2. Thursday 4 November 2004 (morning) 3 hours INSTRUCTIONS TO CANDIDATES c IB MATHEMATICS HIGHER LEVEL PAPER DIPLOMA PROGRAMME PROGRAMME DU DIPLÔME DU BI PROGRAMA DEL DIPLOMA DEL BI N/5/MATHL/HP/ENG/TZ/XX 887 Thursday November (morig) hours INSTRUCTIONS TO CANDIDATES! Do ot

More information

PROCEEDINGS OF THE YEREVAN STATE UNIVERSITY AN ALTERNATIVE MODEL FOR BONUS-MALUS SYSTEM

PROCEEDINGS OF THE YEREVAN STATE UNIVERSITY AN ALTERNATIVE MODEL FOR BONUS-MALUS SYSTEM PROCEEDINGS OF THE YEREVAN STATE UNIVERSITY Physical ad Mathematical Scieces 2015, 1, p. 15 19 M a t h e m a t i c s AN ALTERNATIVE MODEL FOR BONUS-MALUS SYSTEM A. G. GULYAN Chair of Actuarial Mathematics

More information

Lecture 13. Lecturer: Jonathan Kelner Scribe: Jonathan Pines (2009)

Lecture 13. Lecturer: Jonathan Kelner Scribe: Jonathan Pines (2009) 18.409 A Algorithmist s Toolkit October 27, 2009 Lecture 13 Lecturer: Joatha Keler Scribe: Joatha Pies (2009) 1 Outlie Last time, we proved the Bru-Mikowski iequality for boxes. Today we ll go over the

More information

Joint Probability Distributions and Random Samples

Joint Probability Distributions and Random Samples STAT5 Sprig 204 Lecture Notes Chapter 5 February, 204 Joit Probability Distributios ad Radom Samples 5. Joitly Distributed Radom Variables Chapter Overview Joitly distributed rv Joit mass fuctio, margial

More information

PSYCHOLOGICAL STATISTICS

PSYCHOLOGICAL STATISTICS UNIVERSITY OF CALICUT SCHOOL OF DISTANCE EDUCATION B Sc. Cousellig Psychology (0 Adm.) IV SEMESTER COMPLEMENTARY COURSE PSYCHOLOGICAL STATISTICS QUESTION BANK. Iferetial statistics is the brach of statistics

More information

SECTION 1.5 : SUMMATION NOTATION + WORK WITH SEQUENCES

SECTION 1.5 : SUMMATION NOTATION + WORK WITH SEQUENCES SECTION 1.5 : SUMMATION NOTATION + WORK WITH SEQUENCES Read Sectio 1.5 (pages 5 9) Overview I Sectio 1.5 we lear to work with summatio otatio ad formulas. We will also itroduce a brief overview of sequeces,

More information

The second difference is the sequence of differences of the first difference sequence, 2

The second difference is the sequence of differences of the first difference sequence, 2 Differece Equatios I differetial equatios, you look for a fuctio that satisfies ad equatio ivolvig derivatives. I differece equatios, istead of a fuctio of a cotiuous variable (such as time), we look for

More information

Advanced Probability Theory

Advanced Probability Theory Advaced Probability Theory Math5411 HKUST Kai Che (Istructor) Chapter 1. Law of Large Numbers 1.1. σ-algebra, measure, probability space ad radom variables. This sectio lays the ecessary rigorous foudatio

More information

Plug-in martingales for testing exchangeability on-line

Plug-in martingales for testing exchangeability on-line Plug-i martigales for testig exchageability o-lie Valetia Fedorova, Alex Gammerma, Ilia Nouretdiov, ad Vladimir Vovk Computer Learig Research Cetre Royal Holloway, Uiversity of Lodo, UK {valetia,ilia,alex,vovk}@cs.rhul.ac.uk

More information

Key Ideas Section 8-1: Overview hypothesis testing Hypothesis Hypothesis Test Section 8-2: Basics of Hypothesis Testing Null Hypothesis

Key Ideas Section 8-1: Overview hypothesis testing Hypothesis Hypothesis Test Section 8-2: Basics of Hypothesis Testing Null Hypothesis Chapter 8 Key Ideas Hypothesis (Null ad Alterative), Hypothesis Test, Test Statistic, P-value Type I Error, Type II Error, Sigificace Level, Power Sectio 8-1: Overview Cofidece Itervals (Chapter 7) are

More information

A Gentle Introduction to Algorithms: Part II

A Gentle Introduction to Algorithms: Part II A Getle Itroductio to Algorithms: Part II Cotets of Part I:. Merge: (to merge two sorted lists ito a sigle sorted list.) 2. Bubble Sort 3. Merge Sort: 4. The Big-O, Big-Θ, Big-Ω otatios: asymptotic bouds

More information

Annuities Under Random Rates of Interest II By Abraham Zaks. Technion I.I.T. Haifa ISRAEL and Haifa University Haifa ISRAEL.

Annuities Under Random Rates of Interest II By Abraham Zaks. Technion I.I.T. Haifa ISRAEL and Haifa University Haifa ISRAEL. Auities Uder Radom Rates of Iterest II By Abraham Zas Techio I.I.T. Haifa ISRAEL ad Haifa Uiversity Haifa ISRAEL Departmet of Mathematics, Techio - Israel Istitute of Techology, 3000, Haifa, Israel I memory

More information

Z-TEST / Z-STATISTIC: used to test hypotheses about. µ when the population standard deviation is unknown

Z-TEST / Z-STATISTIC: used to test hypotheses about. µ when the population standard deviation is unknown Z-TEST / Z-STATISTIC: used to test hypotheses about µ whe the populatio stadard deviatio is kow ad populatio distributio is ormal or sample size is large T-TEST / T-STATISTIC: used to test hypotheses about

More information

Section 11.3: The Integral Test

Section 11.3: The Integral Test Sectio.3: The Itegral Test Most of the series we have looked at have either diverged or have coverged ad we have bee able to fid what they coverge to. I geeral however, the problem is much more difficult

More information

ARTICLE IN PRESS. Statistics & Probability Letters ( ) A Kolmogorov-type test for monotonicity of regression. Cecile Durot

ARTICLE IN PRESS. Statistics & Probability Letters ( ) A Kolmogorov-type test for monotonicity of regression. Cecile Durot STAPRO 66 pp: - col.fig.: il ED: MG PROD. TYPE: COM PAGN: Usha.N -- SCAN: il Statistics & Probability Letters 2 2 2 2 Abstract A Kolmogorov-type test for mootoicity of regressio Cecile Durot Laboratoire

More information

Approximating Area under a curve with rectangles. To find the area under a curve we approximate the area using rectangles and then use limits to find

Approximating Area under a curve with rectangles. To find the area under a curve we approximate the area using rectangles and then use limits to find 1.8 Approximatig Area uder a curve with rectagles 1.6 To fid the area uder a curve we approximate the area usig rectagles ad the use limits to fid 1.4 the area. Example 1 Suppose we wat to estimate 1.

More information

Normal Distribution.

Normal Distribution. Normal Distributio www.icrf.l Normal distributio I probability theory, the ormal or Gaussia distributio, is a cotiuous probability distributio that is ofte used as a first approimatio to describe realvalued

More information

Department of Computer Science, University of Otago

Department of Computer Science, University of Otago Departmet of Computer Sciece, Uiversity of Otago Techical Report OUCS-2006-09 Permutatios Cotaiig May Patters Authors: M.H. Albert Departmet of Computer Sciece, Uiversity of Otago Micah Colema, Rya Fly

More information

Notes on Hypothesis Testing

Notes on Hypothesis Testing Probability & Statistics Grishpa Notes o Hypothesis Testig A radom sample X = X 1,..., X is observed, with joit pmf/pdf f θ x 1,..., x. The values x = x 1,..., x of X lie i some sample space X. The parameter

More information

Confidence Intervals for One Mean

Confidence Intervals for One Mean Chapter 420 Cofidece Itervals for Oe Mea Itroductio This routie calculates the sample size ecessary to achieve a specified distace from the mea to the cofidece limit(s) at a stated cofidece level for a

More information

Probabilistic Engineering Mechanics. Do Rosenblatt and Nataf isoprobabilistic transformations really differ?

Probabilistic Engineering Mechanics. Do Rosenblatt and Nataf isoprobabilistic transformations really differ? Probabilistic Egieerig Mechaics 4 (009) 577 584 Cotets lists available at ScieceDirect Probabilistic Egieerig Mechaics joural homepage: wwwelseviercom/locate/probegmech Do Roseblatt ad Nataf isoprobabilistic

More information

Confidence Intervals for the Mean of Non-normal Data Class 23, 18.05, Spring 2014 Jeremy Orloff and Jonathan Bloom

Confidence Intervals for the Mean of Non-normal Data Class 23, 18.05, Spring 2014 Jeremy Orloff and Jonathan Bloom Cofidece Itervals for the Mea of No-ormal Data Class 23, 8.05, Sprig 204 Jeremy Orloff ad Joatha Bloom Learig Goals. Be able to derive the formula for coservative ormal cofidece itervals for the proportio

More information

LIMIT DISTRIBUTION FOR THE WEIGHTED RANK CORRELATION COEFFICIENT, r W

LIMIT DISTRIBUTION FOR THE WEIGHTED RANK CORRELATION COEFFICIENT, r W REVSTAT Statistical Joural Volume 4, Number 3, November 2006, 189 200 LIMIT DISTRIBUTION FOR THE WEIGHTED RANK CORRELATION COEFFICIENT, r W Authors: Joaquim F. Pito da Costa Dep. de Matemática Aplicada,

More information

Chapter 14 Nonparametric Statistics

Chapter 14 Nonparametric Statistics Chapter 14 Noparametric Statistics A.K.A. distributio-free statistics! Does ot deped o the populatio fittig ay particular type of distributio (e.g, ormal). Sice these methods make fewer assumptios, they

More information

THE ABRACADABRA PROBLEM

THE ABRACADABRA PROBLEM THE ABRACADABRA PROBLEM FRANCESCO CARAVENNA Abstract. We preset a detailed solutio of Exercise E0.6 i [Wil9]: i a radom sequece of letters, draw idepedetly ad uiformly from the Eglish alphabet, the expected

More information

THE COMPLETENESS OF CONVERGENT SEQUENCES SPACE OF FUZZY NUMBERS. Hee Chan Choi

THE COMPLETENESS OF CONVERGENT SEQUENCES SPACE OF FUZZY NUMBERS. Hee Chan Choi Kagweo-Kyugki Math. Jour. 4 (1996), No. 2, pp. 117 124 THE COMPLETENESS OF CONVERGENT SEQUENCES SPACE OF FUZZY NUMBERS Hee Cha Choi Abstract. I this paper we defie a ew fuzzy metric θ of fuzzy umber sequeces,

More information

1. C. The formula for the confidence interval for a population mean is: x t, which was

1. C. The formula for the confidence interval for a population mean is: x t, which was s 1. C. The formula for the cofidece iterval for a populatio mea is: x t, which was based o the sample Mea. So, x is guarateed to be i the iterval you form.. D. Use the rule : p-value

More information

3. Continuous Random Variables

3. Continuous Random Variables Statistics ad probability: 3-1 3. Cotiuous Radom Variables A cotiuous radom variable is a radom variable which ca take values measured o a cotiuous scale e.g. weights, stregths, times or legths. For ay

More information

Riemann Sums y = f (x)

Riemann Sums y = f (x) Riema Sums Recall that we have previously discussed the area problem I its simplest form we ca state it this way: The Area Problem Let f be a cotiuous, o-egative fuctio o the closed iterval [a, b] Fid

More information

University of California, Los Angeles Department of Statistics. Distributions related to the normal distribution

University of California, Los Angeles Department of Statistics. Distributions related to the normal distribution Uiversity of Califoria, Los Ageles Departmet of Statistics Statistics 100B Istructor: Nicolas Christou Three importat distributios: Distributios related to the ormal distributio Chi-square (χ ) distributio.

More information

Lesson 12. Sequences and Series

Lesson 12. Sequences and Series Retur to List of Lessos Lesso. Sequeces ad Series A ifiite sequece { a, a, a,... a,...} ca be thought of as a list of umbers writte i defiite order ad certai patter. It is usually deoted by { a } =, or

More information

π d i (b i z) (n 1)π )... sin(θ + )

π d i (b i z) (n 1)π )... sin(θ + ) SOME TRIGONOMETRIC IDENTITIES RELATED TO EXACT COVERS Joh Beebee Uiversity of Alaska, Achorage Jauary 18, 1990 Sherma K Stei proves that if si π = k si π b where i the b i are itegers, the are positive

More information

The Euler Totient, the Möbius and the Divisor Functions

The Euler Totient, the Möbius and the Divisor Functions The Euler Totiet, the Möbius ad the Divisor Fuctios Rosica Dieva July 29, 2005 Mout Holyoke College South Hadley, MA 01075 1 Ackowledgemets This work was supported by the Mout Holyoke College fellowship

More information

A Study for the (μ,s) n Relation for Tent Map

A Study for the (μ,s) n Relation for Tent Map Applied Mathematical Scieces, Vol. 8, 04, o. 60, 3009-305 HIKARI Ltd, www.m-hikari.com http://dx.doi.org/0.988/ams.04.4437 A Study for the (μ,s) Relatio for Tet Map Saba Noori Majeed Departmet of Mathematics

More information

A CHARACTERIZATION OF MINIMAL ZERO-SEQUENCES OF INDEX ONE IN FINITE CYCLIC GROUPS

A CHARACTERIZATION OF MINIMAL ZERO-SEQUENCES OF INDEX ONE IN FINITE CYCLIC GROUPS INTEGERS: ELECTRONIC JOURNAL OF COMBINATORIAL NUMBER THEORY 5(1) (2005), #A27 A CHARACTERIZATION OF MINIMAL ZERO-SEQUENCES OF INDEX ONE IN FINITE CYCLIC GROUPS Scott T. Chapma 1 Triity Uiversity, Departmet

More information

Center, Spread, and Shape in Inference: Claims, Caveats, and Insights

Center, Spread, and Shape in Inference: Claims, Caveats, and Insights Ceter, Spread, ad Shape i Iferece: Claims, Caveats, ad Isights Dr. Nacy Pfeig (Uiversity of Pittsburgh) AMATYC November 2008 Prelimiary Activities 1. I would like to produce a iterval estimate for the

More information

THE HEIGHT OF q-binary SEARCH TREES

THE HEIGHT OF q-binary SEARCH TREES THE HEIGHT OF q-binary SEARCH TREES MICHAEL DRMOTA AND HELMUT PRODINGER Abstract. q biary search trees are obtaied from words, equipped with the geometric distributio istead of permutatios. The average

More information

Methods of Evaluating Estimators

Methods of Evaluating Estimators Math 541: Statistical Theory II Istructor: Sogfeg Zheg Methods of Evaluatig Estimators Let X 1, X 2,, X be i.i.d. radom variables, i.e., a radom sample from f(x θ), where θ is ukow. A estimator of θ is

More information

Gregory Carey, 1998 Linear Transformations & Composites - 1. Linear Transformations and Linear Composites

Gregory Carey, 1998 Linear Transformations & Composites - 1. Linear Transformations and Linear Composites Gregory Carey, 1998 Liear Trasformatios & Composites - 1 Liear Trasformatios ad Liear Composites I Liear Trasformatios of Variables Meas ad Stadard Deviatios of Liear Trasformatios A liear trasformatio

More information

Inference on Proportion. Chapter 8 Tests of Statistical Hypotheses. Sampling Distribution of Sample Proportion. Confidence Interval

Inference on Proportion. Chapter 8 Tests of Statistical Hypotheses. Sampling Distribution of Sample Proportion. Confidence Interval Chapter 8 Tests of Statistical Hypotheses 8. Tests about Proportios HT - Iferece o Proportio Parameter: Populatio Proportio p (or π) (Percetage of people has o health isurace) x Statistic: Sample Proportio

More information

1 Introduction to reducing variance in Monte Carlo simulations

1 Introduction to reducing variance in Monte Carlo simulations Copyright c 007 by Karl Sigma 1 Itroductio to reducig variace i Mote Carlo simulatios 11 Review of cofidece itervals for estimatig a mea I statistics, we estimate a uow mea µ = E(X) of a distributio by

More information

Estimating the Mean and Variance of a Normal Distribution

Estimating the Mean and Variance of a Normal Distribution Estimatig the Mea ad Variace of a Normal Distributio Learig Objectives After completig this module, the studet will be able to eplai the value of repeatig eperimets eplai the role of the law of large umbers

More information

4.1 Sigma Notation and Riemann Sums

4.1 Sigma Notation and Riemann Sums 0 the itegral. Sigma Notatio ad Riema Sums Oe strategy for calculatig the area of a regio is to cut the regio ito simple shapes, calculate the area of each simple shape, ad the add these smaller areas

More information

Infinite Sequences and Series

Infinite Sequences and Series CHAPTER 4 Ifiite Sequeces ad Series 4.1. Sequeces A sequece is a ifiite ordered list of umbers, for example the sequece of odd positive itegers: 1, 3, 5, 7, 9, 11, 13, 15, 17, 19, 21, 23, 25, 27, 29...

More information

Determining the sample size

Determining the sample size Determiig the sample size Oe of the most commo questios ay statisticia gets asked is How large a sample size do I eed? Researchers are ofte surprised to fid out that the aswer depeds o a umber of factors

More information

Lecture 10: Hypothesis testing and confidence intervals

Lecture 10: Hypothesis testing and confidence intervals Eco 514: Probability ad Statistics Lecture 10: Hypothesis testig ad cofidece itervals Types of reasoig Deductive reasoig: Start with statemets that are assumed to be true ad use rules of logic to esure

More information

Matrix Transforms of A-statistically Convergent Sequences with Speed

Matrix Transforms of A-statistically Convergent Sequences with Speed Filomat 27:8 2013, 1385 1392 DOI 10.2298/FIL1308385 Published by Faculty of Scieces ad Mathematics, Uiversity of Niš, Serbia vailable at: http://www.pmf.i.ac.rs/filomat Matrix Trasforms of -statistically

More information

Vladimir N. Burkov, Dmitri A. Novikov MODELS AND METHODS OF MULTIPROJECTS MANAGEMENT

Vladimir N. Burkov, Dmitri A. Novikov MODELS AND METHODS OF MULTIPROJECTS MANAGEMENT Keywords: project maagemet, resource allocatio, etwork plaig Vladimir N Burkov, Dmitri A Novikov MODELS AND METHODS OF MULTIPROJECTS MANAGEMENT The paper deals with the problems of resource allocatio betwee

More information

Lecture 7: Borel Sets and Lebesgue Measure

Lecture 7: Borel Sets and Lebesgue Measure EE50: Probability Foudatios for Electrical Egieers July-November 205 Lecture 7: Borel Sets ad Lebesgue Measure Lecturer: Dr. Krisha Jagaatha Scribes: Ravi Kolla, Aseem Sharma, Vishakh Hegde I this lecture,

More information

arxiv:1506.03481v1 [stat.me] 10 Jun 2015

arxiv:1506.03481v1 [stat.me] 10 Jun 2015 BEHAVIOUR OF ABC FOR BIG DATA By Wetao Li ad Paul Fearhead Lacaster Uiversity arxiv:1506.03481v1 [stat.me] 10 Ju 2015 May statistical applicatios ivolve models that it is difficult to evaluate the likelihood,

More information

A Faster Clause-Shortening Algorithm for SAT with No Restriction on Clause Length

A Faster Clause-Shortening Algorithm for SAT with No Restriction on Clause Length Joural o Satisfiability, Boolea Modelig ad Computatio 1 2005) 49-60 A Faster Clause-Shorteig Algorithm for SAT with No Restrictio o Clause Legth Evgey Datsi Alexader Wolpert Departmet of Computer Sciece

More information

Universal coding for classes of sources

Universal coding for classes of sources Coexios module: m46228 Uiversal codig for classes of sources Dever Greee This work is produced by The Coexios Project ad licesed uder the Creative Commos Attributio Licese We have discussed several parametric

More information

Alternatives To Pearson s and Spearman s Correlation Coefficients

Alternatives To Pearson s and Spearman s Correlation Coefficients Alteratives To Pearso s ad Spearma s Correlatio Coefficiets Floreti Smaradache Chair of Math & Scieces Departmet Uiversity of New Mexico Gallup, NM 8730, USA Abstract. This article presets several alteratives

More information

THE TWO-VARIABLE LINEAR REGRESSION MODEL

THE TWO-VARIABLE LINEAR REGRESSION MODEL THE TWO-VARIABLE LINEAR REGRESSION MODEL Herma J. Bieres Pesylvaia State Uiversity April 30, 202. Itroductio Suppose you are a ecoomics or busiess maor i a college close to the beach i the souther part

More information

SUMS OF GENERALIZED HARMONIC SERIES. Michael E. Ho man Department of Mathematics, U. S. Naval Academy, Annapolis, Maryland

SUMS OF GENERALIZED HARMONIC SERIES. Michael E. Ho man Department of Mathematics, U. S. Naval Academy, Annapolis, Maryland #A46 INTEGERS 4 (204) SUMS OF GENERALIZED HARMONIC SERIES Michael E. Ho ma Departmet of Mathematics, U. S. Naval Academy, Aapolis, Marylad meh@usa.edu Courtey Moe Departmet of Mathematics, U. S. Naval

More information