Scholarly article on topic 'The consistency of estimator under fixed design regression model with NQD errors'

The consistency of estimator under fixed design regression model with NQD errors Academic research paper on "Mathematics"

0
0
Share paper
Academic journal
J Inequal Appl
OECD Field of science
Keywords
{""}

Academic research paper on topic "The consistency of estimator under fixed design regression model with NQD errors"

0 Journal of Inequalities and Applications

a SpringerOpen Journal

RESEARCH

Open Access

The consistency of estimator under fixed design regression model with NQD errors

Jianhua Shi1,2*, Xiaoping Chen2,3 and Yong Zhou2

Correspondence: v0085@126.com 1SchoolofMathematicsand Statistics, Minnan NormalUniversity, Zhangzhou, China 2Schoolof Statistics and Management, Shanghai University of Finance and Economics, Shanghai, 200433, China Full list ofauthor information is available at the end ofthe article

Abstract

In this article, based on NQD samples, we investigate the fixed design nonparametric regression model, i.e. Ynk = g(xnk) + enk for 1 < k < n, where enk are pairwise NQD random errors, xnk are fixed design points, and g(-) is an unknown function. The nonparametric weighted estimator gn(■) of g(-) will be introduced and its consistency is studied. As a special case, the consistency result for weighted kernel estimators of the model is established. This extends the earlier work on independent random and dependent random errors to the NQD case.

Keywords: pairwise NQD random sequences; nonparametric regression model; weighted estimator

ft Spri

ringer

1 Introduction

In regression analysis, it is a common practice to investigate the functional relationship between the responses and design points. The nonparametric regression model provides a useful explanatory and diagnostic tool for this purpose. One may see Müller [1] and Hardle [2] for many examples about this and good introductions to the general subject area.

To begin with, consider the fixed design nonparametric regression model

Ynk = g(Xnk) + Snk, 1 < k < n.

Here Xnk, 1 < k < n, are known fixed design points, and enk are random errors. g(-) is an unknown regression function. As an estimate of g(-), we consider the following general linear smoother:

gn(x) = ^ Wnk (x) Ynk,

where the weight functions wnk(x) depend on X, Xn 1,..., Xnn.

It is well known that Georgiev [3] first proposed the estimator above, and the estimator subsequently have been studied by many authors. A brief review of the theoretic development in recent years is worth mentioning. Results on snk being assumed to be independent, consistency and asymptotic normality have been investigated by Georgiev [4] and Müller [5] among others. Results for the case when the enk are dependent have also been

©2014 Shi et al.; licensee Springer. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the originalworkis properly cited.

studied by various authors in recent years. Roussas et al. [6] established asymptotic normality of gn(x) assuming that the errors are from a strictly stationary stochastic process under the strong mixing condition. Tran etal. [7] discussed again asymptotic normality of gn(x) assuming that the errors form a weakly stationary linear process with a martingale difference sequence. Hu et al. [8] gave the mean consistency, complete consistency, and asymptotic normality of regression models based on linear process errors. Under negatively associated sequences, Liang and Jing [9] presented some asymptotic properties for estimates of nonparametric regression models, and Yang etal. [10] generalized part results of Liang and Jing [9] for negatively associated sequences to the case of negatively orthant dependent sequences, and so on.

In this paper, we shall investigate the above nonparametric regression problem under pairwise NQD errors, which means a more general case for sampling.

Definition 1.1 [11] The pair (X, Y) of random variables X and Y is said to be NQD (negatively quadrant dependent), if

A sequence of random variables {Xn, n > 1} is pairwise NQD random variables (NQD in abbreviation), if (Xi, Yj) is NQD for every i = j, i, j = 1,2,____

It can be deduced from Definition 1.1 that

Moreover, it follows that (1.2) also implies (1.1), and hence, (1.1) and (1.2) are actually equivalent.

The definition was introduced by Lehmann [11], which contains independent random variable, NA (negatively associated) random variable and NOD (negatively orthant dependent) random variable et al. as special cases. For the reason of the wide applications of NQD random variables in reliability theory and applications, the notions of NQD random variables have received many concern recently. Some properties of NQD random variables can be found in Lehmann [11], and there is much other relevant literature (e.g. Matula [12], Huang etal. [13], Sung [14], Shi [15], Wang etal. [16], Li and Yang [17]).

However, the pairwise NQD structure is more comprehensive than the NA (negative associated) structure and the NOD (negatively orthant dependent) structure. Concerning the study of the theory of pairwise NQD random variables, due to lack of some key technique tools, such as Bernstain type inequality and exponential inequality etc. still un-established for NQD sequences, investigating the related result is under restraint, especially the estimators of parametric and nonparametric components in regressions model under NQD error's structure. Hence, extending the asymptotic properties of independent and other dependent random variables to the case of NQD variables is highly desirable and of considerable significance in the theory and in applications.

In the article, based on several related lemmas, we investigate the fixed design nonpara-metric regression model with NQD errors. The nonparametric estimator gn(-) ofg(-) will

P(X < x, Y < y) < P(X < x)P(Y < y), Vx,y e R1.

P(X > x, Y > y) < P(X > x)P(Y > y), Vx,y e R1.

be introduced and the usual consistency properties of gn(•), including mean convergence, uniform mean convergence, convergence in probability, etc., are studied under suitable regularity conditions.

The organization of this paper is as follows. In Section 2, we shall present several lemmas for proof of main results, and give the basic assumptions for the nonparametric estimator. We give the further assumption and the main results in Section 3. The proofs of the results will be deferred to Section 4.

2 Some lemmas and basic assumptions 2.1 Some lemmas

We shall begin with a few preliminary lemmas useful in the proofs of our main results. Firstly, a fact about the properties for the NQD random variables is cited from [11].

Lemma 2.1 [11] Let the pair (X, Y) of random variables X and Y be NQD, then:

(1) E(XY) < EX • EY.

(2) P(X > x, Y > y) < P(X > x)P(Y > y), for any x,y e R1.

(3) Iff, g are both non-decreasing (or non-increasing) functions, then f (X) and g(X) are

Lemma 2.2 [18] Let {Xn, n > 1} be a sequence ofpairwise NQD random variables such thatEXn = 0, EXn < to for all n > 1, denote Tj (k) = Ejj X, j > 0, k > 1, then

In the rest below, we assume 0 = xn(o) < x«(i) < x«(2) < ••• < xn(n) = 1 and let Sn = maxi<i<„(x„(i) -xn(k-1)). Furthermore, assume that

(A1 ) K() is bounded and satisfies Lipschitz condition of order a (a >0) on R1, and

£(T(/c))2 ^ EXf,

¿=j+l

fœœ\K(i)| dt < TO; (A2) hn ^ 0 and san /hl+a ^ 0 as n ^to; (A3) /TOTO K(t) dt =l.

Lemma 2.3 If conditions (Al), (A2) hold, then

x e (0,l),

andfor a fixed point t e (0, l/2),

Proof of Lemma 2.3 Denote H (x) = 7(0 < x < 1), where I (■) is the usual indicator function, and

xn(k) xn(k-l)

x xn(k)

Ik (t)| dt

xn(k) xn(k-l)

K |H(Xn(k))

H (t) dt

H(t) dt -

|K (t)|

= Tni(x) + Tn2(x).

We look at each term separately. Note that there is d„(k) e (0,1) (k = 1,2, ...,n) by the mean-value theorem for integrals such that

Mx) •

h--1£

H(xn(k))

x - xn(k) + 9n(k)$n(k)

< h-'J2 M

K x__xW | - k

H(xn(k) - 9n(k)$n(k))

x - xn(k) + 0n(k)Sn,k

|H (xn(k)) 1

x -xn(k) + 9n(k)$n(k) hn

|H(xn(k)) -H(xn(k) - Qn(k)&n(k))

< Mh-1^ ~8n(k)(0n(k)8n(k)lhn)

< M(Snlhn)a lhn,

where 5n(k) • xn(k) -xn(k-l), and we use condition (Al) to the second inequality. Then, according to condition (A2), we conclude

lim |Tnl(x)1 • 0 and lim sup |Tnl(x)1 • 0.

n^œ xe[r ,1-t ]

As for Tn2(x), when x e (0,1), we have

\K (t)\\H(x)-H (x - hnt)\ dt.

Note that by the definition of H(■), limn^TO H(x - hnt)= H(x) for all x e (0,1) and t e R1. Under the integrability of \K(t)|, |Tn2(x)| ^ 0, as n ^to by the dominated convergence theorem, which together with (2.3) implies (2.1).

Again because of /-TO \K(t)\ dt < to and hn ^ 0, if n large sufficiently, one can choose a sufficient small positive number r0, such that when \hnt\ < t0 < t, there is

|K (t)|/( |t|>rolh^dt <-. c2

As a result, for x e [t,1 - t], uniformly

\K(t)\\H(x) - H(x-M|[/(\h„t\ < to) +I(\h„t\ > to)]dt

\K(t)\l(\h„t\ > to) dt < e.

Therefore,

sup \ Tn2(x)\ ^ 0, as n ^to.

xe[z ,1 - t ]

Combining (2.3), then (2.2) holds, as we wanted to show. This completes the proof. □ Lemma 2.4 If conditions (A1), (A2), (A3) hold, then

lim y:

xn(k) xn(k - 1)^1 x xn(k)

= 1, x e (0,1),

andfor a fixed point t e (0,1/2),

lim sup

n^œ xe[T ,1-t ]

xn(k) xn(k- 1) k( x xn(k) . 1

Proof of Lemma 2.4 The proof is similar to those of Lemma 2.3 with \K (i)| replaced by K(t) and using condition (A3), so is omitted here. □

2.2 Basic assumptions

Unless otherwise specified, we assume throughout the paper that the random sample (xnk, Ynk) for 1 < k < n come from the regression model

Ynk = g(xnk) + £nk, 1 < k < n, (.6)

where |enk,1 < k < n} from a sequence of zero mean random errors with the same distribution as |ek,1 < k < n} for each n, {xnk,1 < k < n} are known fixed design points from a compact set A in Rd (d is a positive integer), and g(-) is an unknown real valued regression function and assumed to be bounded on the compact set A.

The present paper investigates the general linear smoother as an estimate of g(-) in the following, defined by

gn (x) = y Wnk (x) Ynk,

where the array of weight functions a)nk(x), 1 < k < n depends on the fixed design points x,xni,...,xnn and on the number of observations n, for which a)nk(x) = 0 for k > n.

In the following section, we denote all continuity points of the function g(•) on set A as C(g). Let the symbol ||x|| be the Euclidean norm of x, M a generic positive constant in the sequel, which could take different values at different places.

3 Main results

We shall establish two different models of consistency for the nonparametric regression estimate gn(x) at a fixed point x. First, we give some assumptions on the weight function o)nk(x) in the following. Similar assumptions on the weighted functions can be found in Georgiev [4], Hu et al. [8], Liang and Jing [9] and Yang et al. [10], etc. We have

(B1) En=1 ^nk(x) — 1, as n — to;

(B2) E^=1 Kk(x) \ < M, Vn;

(B3) En=1 v2nk(x) — 0, as n — to;

(B4) Y^h \Mnk(x)\/(|xnk -x| > a) — 0,as n — to, for a >0.

The weights unk(x), 1 < k < n, in the assumptions are relatively extensive in practice, which can easily be satisfied by the commonly adopted weights used, such as the well-known nearest neighbor weights.

Example 3.1 Let g( • ) be continuous on interval A = [0,1]. Without loss of generality, put xnk = k/n, 1 < k < n. When \xni - x\ = \xnj - x\, assume that \xni -x\ is ahead of \xnj - x\ for x„i < xnj, then a permutation for \xn1 - x\, \xn2 - x\,..., \xnn - x\ can be given as follows:

i ■v(n) ■v i i ■v(n) ■v I . . . i ■v(n) 'V i a* s~ A \AR1(x) < \xR2(x) x\ < < \xRn(x) x r x e A.

Let kn = o(n); we define the nearest neighbor weight as

Unk(x) = k-1^\xnk -x\ < (x) -x^.

Then one can easily verify by the choice of xni and the definition of Ri(x) that conditions (B1)-(B4) are satisfied.

We now state our first result for the mean convergence of gn(x), which, on the opinion of statistics, is asymptotically unbiased of g(x) in the proof of Theorem 3.1.

Theorem 3.1 (Mean convergence) Assume that conditions (B1)-(B4) hold. Let {en, n > 1} be mean zero pairwise NQD sequences with supn>1 Ee"2 < to, if 0 <p < 2, then

lim Egn(x) - g(x)ip = 0, (3.1)

n—>to ' 1

for Vx e C(g).

Another similar form of mean convergence, by using the inequality (^n=1 \ak\P)1/p < (En=1 \ak\a)1/a, 1 < a < P, for any real number sequence {ak,1 < k < n}, is the following.

Theorem 3.1' (Mean convergence) Assume that conditions (B1), (B2), and (B4) hold. Let {en, n > 1} be mean zero pairwise NQD sequences with supn>1 E\en\p < to for some 1 <p < 2, ifYTk=1 \usnk (x)\ — 0, as n — to, with 1 < s < p; then (3.1) holds for Vx e C(g).

For any fixed point x on a compact set A in Rd (d > 1), in order to obtaining uniform convergence for the estimator of g(x), several uniform version of conditions on unk(x) are necessarily replaced by the following:

(B1) SUPxeA IELl ™nk(x) - 1| — 0,as n — to; (B2) supxeA IEn=i 0>nk(x)| < M, Vn; (B3) suPxeA EL1 ^Ik(x) — 0,as n — to;

(B^) supxeA En=1 I^nk(x)|I(l|xnk-x|| > a) — 0,as n — to, for a > 0. Then we are in the position to give the following result.

Theorem 3.2 (Uniform mean convergence) Assume that conditions (Bl)-(B^) hold. Let g(-) be continuous on a compact set A, {en, n > 1}, a mean zero pairwise NQD sequence. If supn>1 Ee2 < to, then

lim supE|gn(x) - g(x)|p = 0, (3.2)

n—TO xeA

for 0 <p < 2.

Remark 3.1 Since the NA sequence and the NOD sequence are an NQD sequence, we generalize some results of Liang and Jing [9] and Yang etal. [10] to the case of NQD errors, respectively. As a consequence, one may get a consistency property for the weighted kernel estimators in the model (2.6).

Corollary 3.1 Assume that conditions (A1), (A2), (A3) hold, and

xn(k) xn(k - 1)

x xn(k) hn

I(|xn(k) - x| > a) — 0, asn — to,for a >0.

Let {en, n > 1} be mean zero pairwise NQD sequences with supn>1 Een < to, g(-) a continuous function on interval (0,1), if 0 <p < 2, then

n—>to

xn(k) xn(k 1) x xn(k)

( ) ( ■ K| -^ I Ynk - g(x)

for Vx e (0,1).

Furthermore, put t e (0,1/2); if (A4) is replaced by

xn(k) xn(k 1)

xn(k) xn(

(A4)' sup ■ ( ) (

xe[T,1 -t] k=1

x xn(k)

i( |x-(-) - x| > a—0,

asn — to, for a >0, then

lim sup E

-—to xe[T ,1-t ]

xn(k) xn(k 1) x xn(k)

Ynk -g(x)

for 0 <p < 2.

Next, we shall give a weak consistency for the estimator of g(x) under the existence of an absolute mean for the variable.

Theorem 3.3 (Convergence in probability) Assume that conditions (Bi), (B2), (B4) hold. Let {en, n > 1} be mean zeropairwise NQD sequences and uniformly bounded by a random variable X in the sense that supn>1 P(|en| > x) < P(\X| > x) for all x > 0. IfE\X\ < to and supk \m„k(x)| = o(1), then

gn (x) ^ g(x), in probability as n ^to, (3.5)

for Vx e C(g).

Corollary 3.2 Assume that conditions (A1)-(A4) hold, and let {en, n > 1} be a mean zero pairwise NQD sequence and uniformly bounded by a random variable X in the sense that supn>1 P(|en |> x) < P(|X| > x) for all x >0, g ( • ) a continuous function on interval (0,1). If EX| < to and Snhn1 = o(1), then

Exn(k) - xn(k-l) „ (x - xn(k)\,, , N , ,

---a I —-- I Ynk ^ g(x), in probability asn ^to,

k=l hn \ hn )

for Vx e (0,1). 4 Proofs for main results

Proof of Theorem 3.1 We write firstly the triangle inequality that

E|gn(x) -g(x) |p < ME{ \gn(x) - Egn(x) |p + |Egn(x) -g(x) \p}. (4.1)

By Jensen's inequality, Lemma 2.2, combine condition (B3) and supn>1 Ee"2 < to, and note that {enk,1 < k < n} has the same distribution as {ek,1 < k < n}; then we have for 0 <p < 2

n \2\ p/2

Kn\" '|P

E\gn(x)-Egn(x)\P < Wnk (X)£kj j

/ n \P/2

< HCw2k(x)Ee2j ^ 0, as n ^to, (4.2)

x e C(g), since {a>nk(x)ek,1 < k < n} are also pairwise NQD sequences according to Lemma 2.1.

Meanwhile, for the bias Egn(x) -g(x), choose a number a >0, and we get the following upper bound:

l-Egn(x) -g(x) j < y\wnk(x)| • ^g(xnk) -g(x)\[/(||xnk -x|| < a)

+/( yxnk - xy > a]+

y^nk (x) - l

\g(x) \, x e C(g).

Because of x e C(g), for any s > 0, there exists a S > 0 such that \g(xnk) -g (x) | < s whenever \\xnk - x\\ < S. Thus, by setting 0 < a < S, conditions (Bi), (B2), (B4) together with the

arbitrariness of e >0 imply that the estimate gn(■) is asymptotically unbiased for g(■), and then

\Egn(x) - g(x) \p ^ 0, n ^to,x e C(g).

Therefore, we can deduce from (4.1), (4.2), (4.3) that (3.1) follows, and this ends the proof. □

Proof of Theorem 3.2 Note that on a compact set A, g(-) is uniformly continuous if it is continuous. Then, similar to the proof of Theorem 3.1, (3.2) can be obtained by the fact that

sup£\gn(x) - g(x)\p < M sup£\gn(x) - Egn(x)\p + sup\£gn(x) - g(x)\p ,

^ed xeA

tending to zero if n — to. This completes the proof. Proof of Corollary 3.1 Note that we have

xn(k) xn(k-l)

x xn (k) hn

< M, Vn,

xn(k) xn(k-1) jyl x xn(k)

iKl^-p^ i, x e (0,1),

by Lemma 2.3 and Lemma 2.4, respectively. And under condition (A2),

n / / \ \ 2

xn(k) - xn(k- 1) kI x - xn(k)

xn(k) xn(k-1) k( x xn(k)

as n — to.

Consequently, according to Theorem 3.1, (3.3) follows.

As to (3.4), similar to above, one may verify, on the interval [t,1 - t], the condition of Theorem 3.2 by the second result of Lemma 2.3 and Lemma 2.4, i.e. (2.2) and (2.5), respectively. This ends the proof. □

Proof of Theorem 3.3 Since x e C(g), by the same reasoning as before, for any positive number e, there is a number S >0, when a e (0, S), such that one may find that |Egn(x) - g(x)| tends to zero by arbitrariness of e >0 and conditions (B1), (B2), (B4). For proving (3.5), note that

\gn (x) - g(x)\ < \gn (x) - £gn(x)\ + \£gn(x) - g(x)\.

We now prove that the random part of r.h.s. in (4.4) tends to zero in probability as n — to. To this end, observe that

\gn (x) - £gn(x)\ =

^Wnk (x)Snk

Next we introduce truncated variables:

Xnl(x) = -I(mnk(x)snk < -l) + on(x)enkl(\rnnk(x)enk\ < l) +I{ù)nk(x)enk > l), Xnk(x) = Onk(x)snkl(\onk(x)£nk \ < l)

Sn(x) = £ onk(x)snk, Sn1 (x) = £X(m(x).

From £\X| < to, it follows that nP{\X\ > n} — 0, as n —^ to. Then for Vx e C(g), when n — to,

P^^x) = Sn(x)) p(\X\ > \&nk(x)\-1) — 0,

since condition (B2) and supk \«nk(x)\ = o(1).

It suffices to show that S^^x) converges to zero in probability for Vx e C(g). Observe that (X^k (x), 1 < k < n} is also a sequence of pairwise NQD random variables by Lemma 2.1, hence by the Chebyshev inequality,

P(\sni>(x)\> e) < e-2£e(x$x))2i

\onk (x)Enk \ > l) + 0^2k (x)EE2kl(\onk (x)Enk \ < l)

k=l k=l

= E (In) + In2),

where the first inequality is due to Lemma 2.2. When we come to In1, we have

In1 < ^ P( Kk (x)X| > 1) < supk |«nk (x)|£ E|X|l( |«nk (x)X| > 1) — 0,

k=1 k=1

as n — to.

Now, choose a number s such that for t > s, P(|X| > t) < et-1. We have

/ tP(|Ek| > t) dt < u 1 / |X| > t) Jo Jo

> t) dt < u lM + E,

which means that m 1 fU tP(\ek\ > t) dt — 0 when u — to. Again

/» /» u /» u

o </ t2 dP(Ek < t) = -u2P(\Ek | > u) + 2/ t^(|Ek |> t)dt < 2 tP(\X\>t)dt. J^Uu Jo Jo

Therefore, it follows that

In2 = y «>lk(x)Ee\l(\wnk(x)sk\ <11

n f Kk(x)|-1

< 2J2 (x) tp( lek |>i)

k=1 J0

« f Kk Mr1

< 2 V w2nk(x) tP(X > t) dt ^ 0, as n ^to. (4.7)

k=1 j0

Hence, Theorem 3.3 follows from (4.4)-(4.7). This ends the proof. □

Proof of Corollary 3.2 By the discussion in Corollary 3.1,itisadirectresult ofTheorem 3.3. This completes the proof of Corollary 3.2. □

Competing interests

The authors declare that they have no competing interests. Authors' contributions

Allauthors read and approved the finalmanuscript.

Author details

1 Schoolof Mathematics and Statistics, Minnan NormalUniversity, Zhangzhou, China. 2Schoolof Statistics and Management, Shanghai University of Finance and Economics, Shanghai, 200433, China. 3Schoolof Mathematics and Computer Science, Fujian NormalUniversity, Fuzhou, China. 4Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing, China.

Acknowledgements

The work was partially supported by NationalNaturalScience Foundation of China (NSFC) (No. 71271128, No. 11301473), the State Key Program of NationalNaturalScience Foundation ofChina (71331006), Science Fund forCreative Research Groups, China (11021161), NCMIS, Graduate Innovation Foundation of Shanghai University of Finance and Economics, China (CXJJ2012-423, CXJJ2013-451), and NaturalScience Foundation of Fujian Province, China (2012J01028). The authors wish to express their heartfelt thanks to two anonymous referees for their carefulreading of the manuscript and helpful suggestions.

Received: 21 September 2013 Accepted: 10 February 2014 Published: 25 Feb 2014 References

1. Müller, HG: Nonparametric Analysis of LongitudinalData. Springer, Berlin (1988)

2. Hardle, W: Applied Nonparametric Regression. Cambridge University Press, New York (1990)

3. Georgiev, AA: Localproperties of function fitting estimates with applications to system identification. In: MathematicalStatistics and Applications. Proceedings of the 4th Pannonian Symposium on MathematicalStatistics, Bad Tatzmannsdorf, Austria, 4-10 September 1983, pp. 141-151. Reidel, Dordrecht (1985)

4. Georgiev, AA: Consistent nonparametric multiple regression: the fixed design case. J. Multivar. Anal. 25(1), 100-110 (1988)

5. Müller, HG: Weak and universalconsistency of moving weighted averages. Period. Math. Hung. 18(3), 241-250 (1987)

6. Roussas, GG, Tran, LT, Ioannides, DA: Fixed design regression for time series: asymptotic normality. J. Multivar. Anal. 40(2), 262-291 (1992)

7. Tran, L, Roussas, G, Yakowitz, S, Van, BT: Fixed-design regression for linear time series. Ann. Stat. 24(3), 975-991 (1996)

8. Hu, SH, Pan, GM, Gao, QB: Estimation problems for a regression modelwith linear process errors. Appl. Math. J. Chin. Univ. Ser. A 18(1), 81-90 (2003)

9. Liang, HY, Jing, BY: Asymptotic properties for estimates of nonparametric regression models based on negatively associated sequences. J. Multivar. Anal. 95(2), 227-245 (2005)

10. Yang, WZ, Wang, XJ, Wang, XH, Hu, SH: The consistency for estimator of nonparametric regression modelbased on NOD errors. J. Inequal. Appl. 2012,140 (2012)

11. Lehmann, EL: Some concepts of dependence. Ann. Math. Stat. 43(3), 1137-1153 (1966)

12. Matula, P: A note on the almost sure convergence of sums of negatively dependent random variables. Stat. Probab. Lett. 15(3), 209-213(1992)

13. Huang, HW, Wang, DC, Wu, QY, Zhang, QX: A note on the complete convergence for sequences of pairwise NQD random variables. J. Inequal. Appl. 2011, 92 (2011)

14. Sung, SH: Convergence in r-mean of weighted sums of NQD random variables. Appl. Math. Lett. 26(1), 18-24 (2013)

Page 12 of 12

15. Shi, JH: On the strong law of large numbers for pairwise NQD random variables with different distributions. Acta Math. Appl. Sin. 34(1), 122-130 (2011)

16. Wang, YB, Su, C, Liu, XG: On some limit properties for pairwise NQD sequences. Acta Math. Appl. Sin. 21 (3), 404-414 (1998)

17. Li, R, Yang, WG: Strong convergence of pairwise NQD random sequences. J. Math. Anal. Appl. 344(2), 741-747 (2008)

18. Wu, QY: Convergence properties of pairwise NQD sequences. Acta Math. Sin. 45(3), 617-624 (2002)

10.1186/1029-242X-2014-92

Cite this article as: Shi et al.: The consistency of estimator under fixed design regression model with NQD errors.

Journal of Inequalities and Applications 2014, 2014:92

Submit your manuscript to a SpringerOpen journal and benefit from:

► Convenient online submission

► Rigorous peer review

► Immediate publication on acceptance

► Open access: articles freely available online

► High visibility within the field

► Retaining the copyright to your article

Submit your next manuscript at ► springeropen.com