Estimation of Average Paddy Production of Pira Nagar Village at Barabanki District in India

S. K. Yadav*, Shanya Baghel, Sugandha Saxena and Avinash Kumar Singh

Department of Statistics, Babasaheb Bhimrao Ambedkar University, Lucknow, India

E Mail: drskystats@gmail.com; baghel.shanya@gmail.com; saxenasugandha0911@gmail.com; avinashsingh543@gmail.com

Corresponding Author

## Abstract

The idea of the present paper is the use of the known information on study variable for enhanced estimation of average paddy production of Pira Nagar village at Barabanki District in India under the Simple Random Sampling Scheme. This known information is utilzed in the form of median of primary variable as it is readily available and does not require every unit of the population to be inquired. The Bias and MSE of the suggested estimator are derived up to approximation of degree one.The minimum value of the MSE of suggested estimator is also obtained by optimizing the characterizing scalar. The MSE has also been compared with the considered competing estimators both theoretically and empirically. The theoretical efficiency conditions of the suggested estimator to be better than the considered estimators are verified using natural population on primary data collected from Pira Nagar Village at Barabanki District of Uttar Pradesh state in India.

Keywords: Study variable, median of study variable, simple random sampling, bias, MSE, PRE

## 1 Introduction

Cochran (1940) was the first who used the auxiliary information for enhancing the efficiency of the estimator of population mean. Since then a number of researchers have modified the usual ratio estimator, utilizing the auxiliary information in the form of its various parameters. Taking data on auxiliary variable requires extra cost. Sometimes it is not feasible to bear the extra cost and sometimes it becomes impossible to get information on auxiliary variable. In such situations we need to find an alternative to the auxiliary information. One of the solutions for such situations is to obtain some easily accessible chracteristics on study variable itself as supplementary information. For example Let the study variable is monthly salary of the workers working at a place. Most of the workers are unwilling to reveal their exact salary or in case of free lancers they don’t have a fixed salary. But they can tell whether their salary lies between 10000–15000, 15000–20000 and so on. Hence it is easier for them to tell the range within which their earning lies. In such kind of situations median can be obtained for the whole data and can be used for elevated estimation of average salary of the workers. In the present investigation, we make use of the known median of the main variable to obtain the estimate of average production of paddy crop at Pira Nagar Village of Barabanki District of Uttar Prdesh state in India.

Let $U=U1,U2,…,UN$ be population containing N units which are distinct and may be identified. The problem is to estimate the population mean $Y¯=1N⁢∑i=1NYi$ of the main variable Y with higher efficiency. The most suitable estimator for $Y¯$ is the sample mean $y¯$. When it costs high to get auxiliary information, we may consider additional information on Y and may suggest a modified ratio type estimator for enhanced estimation of $Y¯$. If we go for biased estimator, we can obtain much lesser MSE than the variance of $y¯$ and MSE/Variance of other existing biased and unbiased estimators of $Y¯$.

### 1.1 Notation

 N: Population Size n: Sample Size f: $nN$: Sampling Fraction $𝑪𝒏𝑵$: All possible samples of size n Y: Study variable M: Median of the Y X: Auxiliary variable $𝒀¯,𝑿¯$: Population means $𝒚¯,𝒙¯$: Sample means $𝝆$: Correlation coefficient between X and Y $𝜷$: Regression coefficient of Y on X $𝜷𝟏$: Coefficient of Skewness of X $𝜷𝟐$: Coefficient of Kurtosis of X $𝑴¯$: Average of sample medians of Y m: Sample median of Y $𝑸𝒓$: Interquartile range B($⋅$): Bias of the estimator V($⋅$): Variance of the estimator MSE($⋅$): Mean squared error of the estimator $𝑸𝟏,𝑸𝟐,𝑸𝟑$: Quartiles of X $𝑪𝒚$, $Cx$, $𝑪𝒎$: Coefficient of variation of x, y and m respectively $𝑪𝒚⁢𝒙$, $Cy⁢m$: Relative Covariances $λ:1-fn$ PRE(e,p): Percentage relative efficiency of the proposed $M⁢S⁢E⁢(e)M⁢S⁢E⁢(p)×100$: estimator(p) with respect to the existing estimator (e)

### 1.2 Formulae

Variance of Study Variable:

 $𝑽⁢(𝒚¯)=𝟏(𝑵𝒏)⁢∑𝒊=𝟏(𝑵𝒏)(𝒚𝒊-𝒀¯)𝟐=𝟏-𝒇𝒏⁢𝑺𝒚𝟐$

Variance of Auxiliary variable:

 $𝑽⁢(𝒙¯)=𝟏(𝑵𝒏)⁢∑𝒊=𝟏(𝑵𝒏)(𝒙𝒊-𝑿¯)𝟐=𝟏-𝒇𝒏⁢𝑺𝒙𝟐$

where

 $Sy2=1N-1⁢∑i=1N(yi-Y¯)2 and Sx2=1N-1⁢∑i=1N(xi-X¯)2$

Mean of medians of possible samples

 $M¯=1(Nn)⁢∑i=1(Nn)mi$

Variance of Sample Median of Y

 $V⁢(m)=1(Nn)⁢∑i=1(Nn)(mi-M)2$

Covariance of $𝒚¯$ and $𝒙¯$

 $C⁢o⁢v⁢(y¯,x¯)=1(Nn)⁢∑i=1(Nn)(yi-Y¯)⁢(xi-X¯)=1-fn⁢Sy⁢x$

where

 $Sy⁢x=1N-1⁢∑i=1N(yi-Y¯)⁢(xi-X¯)$

Covariance of $𝒚¯$ and $𝒎$

 $Cov⁢(y¯,m)=1(Nn)⁢∑i=1(Nn)(yi-Y¯)⁢(mi-M)$

Coefficient of Variations

 $Cx⁢x$ $=V⁢(x¯)X¯2=Cx2$ $Cy⁢y$ $=V⁢(y¯)Y¯2=Cy2$ $Cm⁢m$ $=V⁢(m)M2=Cm2$ $Cy⁢m$ $=C⁢o⁢v⁢(y¯,m)M⁢Y¯$ $Cy⁢x$ $=C⁢o⁢v⁢(y¯,x¯)Y¯⁢X¯$

## 2 Literature Review of Existing Estimators

Under this section, various estimators of $Y¯$ along with their biases and MSEs are presented in Table 1. It is well known that in simple random sampling technique ($y¯$, $x¯$) are unbiased estimators for ($Y¯$, $X¯$) respectively.

Table 1 Various existing estimators of $𝒀¯$ along with their biases & MSEs

 S.No. Estimators Bias MSE/Variance 1 $t0=y¯=1n⁢∑i=1nyi$ Sample Mean – $1-fn⁢Y¯2⁢Cy2$ 2 $Y¯^l⁢r=y¯+β⁢(X¯-x¯)$ Watson(1937) – $1-fn⁢Y¯2⁢Cy2⁢(1-ρ2)$ 3 $t1=y¯⁢(X¯x¯)$ Cochron(1940) $1-fn⁢Y¯⁢(Cx2-ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+Cx2-2⁢ρ⁢Cy⁢Cx)$ 4 $t2=X¯⁢r¯$ $t2′=X¯⁢r¯+n⁢(N-1)N⁢(n-1)⁢(y¯-r¯⁢x¯)$ Goodman(1958) – $1-fn⁢Y¯2⁢(Cy2+Cx2-2⁢ρ⁢Cy⁢Cx)$ 5 $t3=(1-α)⁢y¯+α⁢y¯⁢X¯x¯$ Chakrabarty (1979) $1-fn⁢Y¯⁢(α2⁢Cx2-α⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+α2⁢Cx2-2⁢α⁢ρ⁢Cy⁢Cx)$ 6 $t4=y¯⁢{2-(x¯X¯)w}$ Sahai (1980) $1-fn⁢Y¯⁢(-w⁢(1-w)2⁢Cx2-w⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+w2⁢Cx2-2⁢w⁢ρ⁢Cy⁢Cx)$ 7 $t5=y¯⁢(X¯+Cxx¯+Cx)$ Sisodia (1981) $1-fn⁢Y¯⁢(R52⁢Cx2-R5⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+R52⁢Cx2-2⁢R5⁢ρ⁢Cy⁢Cx)$ 8 $t6=y¯⁢exp⁢(X¯-x¯X¯+x¯)$ Tuteja(1991) $1-f8⁢n⁢Y¯⁢(3⁢Cx2-4⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+Cx24-ρ⁢Cy⁢Cx)$ 10 $t8=y¯⁢exp⁢(X¯⁢β2-Cxx¯⁢β2+Cx)$ Upadhyaya(1999) $1-fn⁢Y¯⁢(R82⁢Cx2-R8⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+R82⁢Cx2-2⁢R8⁢ρ⁢Cy⁢Cx)$ 11 $t9=y¯⁢(X¯2x¯2)$ Kadilar(2003) $1-fn⁢Y¯⁢(3⁢Cx2-2⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+4⁢Cx2-4⁢ρ⁢Cy⁢Cx)$ 12 $t10=y¯⁢(X¯⁢β1+Sxx¯⁢β1+Sx)$ Singh(2003) $1-fn⁢Y¯⁢(R(10)2⁢Cx2-R(10)⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+R(10)2⁢Cx2-2⁢R(10)⁢ρ⁢Cy⁢Cx)$ 13 $t11=y¯⁢(X¯+ρx¯+ρ)$ Singh (2003) $1-fn⁢Y¯⁢(R(11)2⁢Cx2-R(11)⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+R(10)2⁢Cx2-2⁢R(10)⁢ρ⁢Cy⁢Cx)$ 14 $t12=y¯⁢(X¯+β2x¯+β2)$ Singh et al. $1-fn⁢Y¯⁢(R(11)2⁢Cx2-R(11)⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+R(12)2⁢Cx2-2⁢R(12)⁢ρ⁢Cy⁢Cx)$ 15 $t13=y¯⁢(X¯+β1x¯+β1)$ Yan and Tian (2010) $1-fn⁢Y¯⁢(R(13)2⁢Cx2-R(13)⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+R(13)2⁢Cx2-2⁢R(13)⁢ρ⁢Cy⁢Cx)$ 16 $t14=y¯⁢(X¯⁢β1+β2x¯⁢β1+β2)$ Yan and Tian (2010) $1-fn⁢Y¯⁢(R(14)2⁢Cx2-R(14)⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+R(14)2⁢Cx2-2⁢R(14)⁢ρ⁢Cy⁢Cx)$ 17 $t15=y¯⁢(X¯⁢Cx+β1x¯⁢Cx+β1)$ Yan and Tian (2010) $1-fn⁢Y¯⁢(R(15)2⁢Cx2-R(15)⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+R(15)2⁢Cx2-2⁢R(15)⁢ρ⁢Cy⁢Cx)$ 18 $t16=y¯⁢(X¯⁢β2+β1x¯⁢β2+β1)$ Yan and Tian (2010) $1-fn⁢Y¯⁢(R(16)2⁢Cx2-R(16)⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+R(16)2⁢Cx2-2⁢R(16)⁢ρ⁢Cy⁢Cx)$ 19 $t17=y¯⁢X¯x¯⁢(1-k⁢x¯⁢s2n⁢x¯3)-1$ Pandey (2011) $1-fn⁢Y¯⁢(k⁢Cx2-ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+k2⁢Cx2-2⁢k⁢ρ⁢Cy⁢Cx)$ 20 $t18=y¯⁢(X¯+Q3x¯+Q3)$ Al-Omari (2012) $1-fn⁢Y¯⁢(R(18)2⁢Cx2-R(18)⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+R(18)2⁢Cx2-2⁢R(18)⁢ρ⁢Cy⁢Cx)$ 21 $t19=y¯⁢(X¯+Qrx¯+Qr)$ Al-Omari (2012) $1-fn⁢Y¯⁢(R(19)2⁢Cx2-R(19)⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+R(19)2⁢Cx2-2⁢R(19)⁢ρ⁢Cy⁢Cx)$ 22 $t20=y¯⁢(X¯+Mdx¯+Md)$ Subramani and Kumarpandiyan (2012) $1-fn⁢Y¯⁢(R(20)2⁢Cx2-R(20)⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+R(20)2⁢Cx2-2⁢R(20)⁢ρ⁢Cy⁢Cx)$ 23 $t21=y¯⁢(X¯⁢Cx+Mdx¯⁢Cx+Md)$ Subramani and Kumarpandiyan (2012) $1-fn⁢Y¯⁢(R(21)2⁢Cx2-R(21)⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+R(21)2⁢Cx2-2⁢R(21)⁢ρ⁢Cy⁢Cx)$ 24 $t22=y¯⁢(X¯x¯)1/2$ Swain (2014) $1-f8⁢n⁢Y¯⁢(3⁢Cx2-4⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+Cx24-ρ⁢Cy⁢Cx)$ 25 $t23=α′+tR⁢e+(1-α′)⁢tP⁢e$ Yadav and Mishra (2015) $Y¯⁢[18⁢(4⁢α-1-7⁢f+4⁢f2)⁢Cx2-(2⁢α-1)2⁢(1-f)⁢ρ⁢Cy⁢Cx]$ $Y¯2⁢(1-f)n⁢(Cy2+α2⁢Cx24-α⁢ρ⁢Cy⁢Cx)$ 26 $t24=y¯⁢(X¯+nx¯+n)$ Jerjuddin and Kishun (2016) $1-fn⁢Y¯⁢(R(24)2⁢Cx2-R(24)⁢ρ⁢Cy⁢Cx)$ $1-fn⁢Y¯2⁢(Cy2+R(24)2⁢Cx2-2⁢R(24)⁢ρ⁢Cy⁢Cx)$ 27 $t25=y¯⁢(X¯+Cxx¯+Cx)b1$ Soponviwatkul and Lawson (2017) $1-fn⁢Y¯⁢[b1⁢(b1+1)2⁢R52⁢Cx2-b1⁢R5⁢ρ⁢Cy⁢Cx]$ $1-fn⁢Y¯2⁢Cy2⁢(1-ρ2)$ 28 $t26=y¯⁢(X¯+ρx¯+ρ)b2$ Soponviwatkul and Lawson (2017) $1-fn⁢Y¯⁢[b2⁢(b2+1)2⁢R112⁢Cx2-b2⁢R11⁢ρ⁢Cy⁢Cx]$ $1-fn⁢Y¯2⁢Cy2⁢(1-ρ2)$ 29 $t27=ω1⁢y¯+(1-ω1)⁢(y¯⁢X¯x¯)$ Ijaz and Ali (2018) $1-fn⁢Y¯⁢ρ⁢Cy⁢(Cx-ρ)$ $1-fn⁢Y¯2⁢Cy2⁢(1-ρ2)$ 30 $t28=ω2⁢y¯+(1-ω2)⁢(y¯⁢exp⁢X¯-x¯x¯-X¯)$ Ijaz and Ali (2018) $1-fn⁢Y¯⁢ρ⁢Cy⁢(14⁢Cx-ρ⁢Cy)$ $1-fn⁢Y¯2⁢Cy2⁢(1-ρ2)$ 31 $t29=y¯⁢(a⁢b⁢X¯+c⁢da⁢b⁢x¯+c⁢d)$ Yadav et al. (2019) – $1-fn⁢Y¯2⁢(Cy2-Cy⁢x2Cx2)$

Note: We will denote our measures in terms of Coefficient of Variation because it is a relative measure and most suitable to compare two series.

where,

 $R5$ $=X¯X¯+Cx,R7=X¯⁢CxX¯⁢Cx+β2,R8=X¯⁢β2X¯⁢β2+Cx,R10=X¯⁢β1X¯⁢β1+Sx,$ $R11$ $=X¯X¯+ρ,R12=X¯X¯+β2,R13=X¯X¯+β1,R14=X¯⁢β1X¯⁢β1+β2,$ $R15$ $=X¯⁢CxX¯⁢Cx+β1,R16=X¯⁢β2X¯⁢β2+β1,k=ρ⁢CyCx,R18=X¯X¯+Q3,$ $R19$ $=X¯X¯+Qr,R20=X¯X¯+Md,R21=X¯⁢CxX¯⁢Cx+Md,α′=ρ4⁢CyCx,$ $ω1$ $=w/R5,ω2=w/R11,w=ρ⁢CyCx$

a,b,c,d = Constants or Parametric Value

## 3 Proposed Estimator Based on Median

If the median M of Y is known so by utilizing it, we may suggest an elevated estimator of $Y¯$ as,

 $𝒕=𝒚¯+𝜶⁢log⁢𝒎𝑴$ (1)

Where $α$ is chosen such that MSE(t) is minimum. The bias and MSE of t up to approximation of order one is given by,

 $B⁢(t)$ $=α⁢B⁢(m)M-α2⁢λ⁢Cm2$ (2) $𝑀𝑆𝐸⁢(t)$ $=λ⁢Y¯2⁢Cy2+λ⁢α2⁢Cm2+2⁢a⁢Y¯⁢λ⁢Cym$ (3)

where $λ=1-fn$

Hence

 $αm⁢i⁢n$ $=-Y¯⁢CymCm2$ (4) $and⁢M⁢S⁢Em⁢i⁢n⁢(t)$ $=λ⁢Y¯2⁢(Cy2-CYm2Cm2)$ (5)

### 3.1 Proposed Estimator Based on Median: Detailed Study

Let

 $e0$ $=y¯-Y¯Y¯ and e1=m-MM$ $then⁢y¯$ $=Y¯⁢(1+e0) and m=M⁢(1+e1)$

So,

 $E⁢(e0)$ $=E⁢(y¯-Y¯Y¯)=E⁢(y¯)-Y¯Y¯=0 or E⁢(e0)=0$ (6) $E⁢(e1)$ $=E⁢(m-MM)=E⁢(m)-MM or$ $E⁢(e1)$ $=M¯-MM=B⁢(m)M$ (7) $E⁢(e02)$ $=E⁢(y¯-Y¯Y¯)2=E⁢(y¯-Y¯)2Y¯2=V⁢(y¯)Y¯2=1-fn⁢Cy2$ $E⁢(e02)$ $=λ⁢Cy2$ (8) $E⁢(e12)$ $=E⁢(m-MM)2=E⁢(m-M)2M2=V⁢(m)M2=1-fn⁢Cm2$ $E⁢(e12)$ $=λ⁢Cm2$ (9) $E⁢(e0⁢e1)$ $=E⁢[((y¯-Y¯)Y¯)⁢(m-MM)]=C⁢o⁢v⁢(y¯,m)Y¯⁢M=λ⁢Cym$ $E⁢(e0⁢e1)$ $=Cym$ (10)

Hence the Estimator can be rewritten as,

 $t=Y¯+Y¯⁢e0+α=(e1-e122)$ (11)

For Biasedness,

 $t$ $=Y¯+Y¯⁢e0+a⁢(e1-e122)$ $(t-Y¯)$ $=Y¯⁢e0+a⁢e1-a⁢e122$ $E⁢(t-Y¯)$ $=Y¯⁢E⁢(e0)+a⁢E⁢(e1)-a2⁢E⁢(e12)$ $B⁢(t)$ $=0+a⁢B⁢(m)M-a2⁢λ⁢Cm2$

(from Equation (6), (3.1) and (8))

Hence Biasedness is :

 $𝑩⁢(𝒕)=𝜶⁢𝑩⁢(𝒎)𝑴-𝜶𝟐⁢𝝀⁢𝑪𝒎𝟐$ (12)

For MSE,

 $𝑀𝑆𝐸⁢(t)$ $=E⁢(t-Y¯)2$ $=E⁢(Y¯⁢e0+α⁢e1)2$ (second and higher order terms are ignored) $=E⁢(Y¯2⁢e02+α2⁢e12+2⁢Y¯⁢α⁢e0⁢e1)$ $=Y¯2⁢E⁢(e02)+α⁢E⁢(e12)+2⁢α⁢E⁢(e0⁢e1)$

Hence,

 $𝑴⁢𝑺⁢𝑬⁢(𝒕)=𝝀⁢𝒀¯𝟐⁢𝑪𝒚𝟐+𝝀⁢𝜶𝟐⁢𝑪𝒎𝟐+𝟐⁢𝜶⁢𝒀¯⁢𝝀⁢𝑪𝒚𝒎$ (13)

(from equation (8), (9) and (10))

Minimum Value of $𝒂$

For Minimum value of $α$, we should have

 $∂⁡𝑀𝑆𝐸⁢(t)∂⁡a=0 and ∂2⁡𝑀𝑆𝐸⁢(t)∂⁡α2>0$

So,

 $∂⁡𝑀𝑆𝐸⁢(t)∂⁡α$ $=02⁢α⁢λ⁢Cm2+2⁢Y¯⁢λ⁢Cym=0$ $𝜶𝒎⁢𝒊⁢𝒏$ $=-𝒀¯⁢𝑪𝒚𝒎𝑪𝒎𝟐$ (14)

For Minima,

 $∂2⁡𝑀𝑆𝐸⁢(t)∂⁡α2=2⁢λ⁢Cm2>0$

Hence $a$ has a minimum value.

Minimum Value of MSE

Hence Minimum value of MSE(t) is obtained by putting the Value of $α$ in Equation (3.3)

 $M⁢S⁢Em⁢i⁢n⁢(t)$ $=λ⁢Y¯2⁢Cy2+λ⁢Y¯2⁢Cym2⁢Cm2Cm4-2⁢Y¯⁢CymCm2⁢Y¯⁢λ⁢Cym$ $=λ⁢Y¯2⁢Cy2+λ⁢Y¯2⁢Cym2Cm2-2⁢λ⁢Y¯2⁢Cym2Cm2$ $𝑴⁢𝑺⁢𝑬𝒎⁢𝒊⁢𝒏⁢(𝒕)$ $=𝝀⁢𝒀¯𝟐⁢(𝑪𝒚𝟐-𝑪𝒚𝒎𝟐𝑪𝒎𝟐)$ (15)

## 4 Efficiency Comparison

In this section the suggested estimator is being compared with the competing estimators of $Y¯$ and the efficiency conditions are presented in Table 2.

Table 2 Efficiency comparison

 S.No. $𝑀𝑆𝐸⁢(t)<𝑀𝑆𝐸⁢(⋅)$ Condition 1 $𝑀𝑆𝐸⁢(t)≤V⁢(y¯)$ $Cym2Cm2≥0$ 2 $𝑀𝑆𝐸⁢(t)≤𝑀𝑆𝐸⁢(t2′)$ $Cym2Cm2≥Cx⁢(2⁢ρ⁢Cy-Cx)$ 3 $𝑀𝑆𝐸⁢(t)≤𝑀𝑆𝐸⁢(tj),𝑀𝑆𝐸⁢(y¯l⁢r)$ $CymCm2≥ρ2⁢Cy2;j=3,4,25,…,29$ 4 $𝑀𝑆𝐸⁢(t)≤𝑀𝑆𝐸⁢(t5)$ $Cym2Cm2≥2⁢R5⁢Cx⁢(2⁢ρ⁢Cy-R52⁢Cx)$ 5 $𝑀𝑆𝐸⁢(t)≤𝑀𝑆𝐸⁢(t6)$ $Cym2Cm2≥Cx⁢(ρ⁢Cy-Cx4)$ 6 $𝑀𝑆𝐸⁢(t)≤𝑀𝑆𝐸⁢(tj)$ $Cym2Cm2≥2⁢Rj⁢Cx⁢(2⁢ρ⁢Cy-Rj2⁢Cx)$; $j=7,8,10⁢…,16,18,…,21$ 7 $𝑀𝑆𝐸⁢(t)≤𝑀𝑆𝐸⁢(t9)$ $Cym2Cm2≥4⁢Cx⁢(ρ⁢Cy-Cx)$ 8 $𝑀𝑆𝐸⁢(t)≤𝑀𝑆𝐸⁢(t17)$ $CymCm2≥ρ2⁢Cy2$ 9 $𝑀𝑆𝐸⁢(t)≤𝑀𝑆𝐸⁢(t22)$ $Cym2Cm2≥Cx⁢(ρ⁢Cy-Cx4)$ 10 $𝑀𝑆𝐸⁢(t)≤𝑀𝑆𝐸⁢(t23)$ $Cym2Cm2≥ρ⁢Cy⁢Cx$ 11 $𝑀𝑆𝐸⁢(t)≤𝑀𝑆𝐸⁢(t24)$ $Cym2Cm2≥2⁢R24⁢Cx⁢(ρ⁢Cy-R242⁢Cx)$

## 5 Numerical Study

Under this section the efficiency conditions of the suggested estimator over competing estimators are verified using real data sets.

### 5.1 Data Collection

To verify the results we have obtained for the Paddy Production data from the Pira Nagar Village of Barabanki Distrcit.The details of the obtained data is as follows:

 Village Pira Nagar District Barabanki Time March 2018 Production Paddy Production Type of Data Primary Data Information Taken Name of Resident Their Area of Cultivation (Unit in Hectares) Yield obtained for each area ( Unit in Quintals): one Quintal=100 Kilogram

For our Numerical Justification we have taken:

 Study Variable Yield denoted as Y Auxiliary Variable Area of Cultivation denoted as X Population Size 52 Sample Size 3

### 5.2 Population Parameters

 Parameter Value Units in N 52 – n 3 – $CnN$ 22100 – $Y¯$ 14.721 Quintal $X¯$ 0.46227 Hectare M 10 Quintal $ρ$ 0.8046229 – $Sy2$ 190.8668 Quintal $Sx2$ 0.15675383 Hectare $Sy⁢x2$ 4.401155 – $Cy2$ 0.8807379 – $Cx2$ 0.7335474 – $Cy⁢x$ 4.401155 – $β1$ 8.1028894 – $β2$ 14.146291 – $β$ 28.0769 – $Q1$ 0.4040 Hectare $Q3$ 0.5050 Hectare $Qr$ 0.2525 Hectare $M¯$ 12.0119 Quintal V(m) 37.7429 Quintal $Cy⁢m2$ 0.2536 – $Cm2$ 0.3774 –

### 5.3 Measurement on Population Parameters

On the basis of the data, The numerical values related to proposed estimators are obtained as follows:

 t 14.0222 B(t) 1.59621 $am⁢i⁢n$ 9.89212 $𝑀𝑆𝐸m⁢i⁢n⁢(t)$ 14.2111

### 5.4 Numerical Comparison

We have constructed a table for numerical comparison of the suggested estimator with the estimators in competition. The following table gives:

• The values of the suggested and competing estimators on the Basis of the data obtained and sample taken.

• The Biases of competing and suggested Estimators.

• Mean Square Error (MSE) of competing and Proposed Estimators.

• The percentage relative efficiencies (PRE) of the suggested over competing estimators of $Y¯$.

Table 4 Biases, MSE’s and respective PRE’s

 Estimator Bias MSE PRE $t0$ 0 $V⁢(t0)=$59.951751 421.866 $y¯l⁢r$ 0 $V⁢(y¯l⁢r)$=21.137907 148.74245 $t1$ 0.40139256 21.837168 153.66298 $t2`$ 0 21.837168 153.66298 $t3$ $-$1.1203056 21.137907 148.74245 $t4$ $-$2.3629292 21.137907 148.74245 $t5$ $-$0.63149751 35.223464 247.8592 $t6$ $-$0.22328973 28.411333 199.92382 $t7$ $-$0.078904053 57.591622 405.25862 $t8$ 0.0076015701 21.138228 148.74471 $t9$ 4.1946732 83.587598 588.18614 $t10$ 0.06976912 21.163736 148.9242 $t11$ $-$0.63958709 34.472784 242.5768 $t12$ $-$0.091234028 57.215612 402.6127 $t13$ $-$0.15151962 55.345217 389.4512 $t14$ $-$0.47740469 43.707413 307.5587 $t15$ $-$0.086059435 57.373669 403.7249 $t16$ $-$0.65903753 30.58859 215.2447 $t17$ 0.27660967 21.493713 151.2462 $t18$ $-$0.65448695 29.277583 206.0194 $t19$ $-$0.51534035 23.893597 168.13359 $t20$ $-$0.62993932 27.185932 191.30097 $t21$ $-$0.64921611 28.605068 201.2871 $t22$ $-$0.22328973 28.411333 199.9238 $t23$ 1.7460142 15.928206 112.0830 $t24$ $-$0.33881397 49.086159 345.4077 $t25$ $-$0.7940365 21.137907 148.74245 $t26$ $-$0.8804347 21.137907 148.74245 $t27$ 0.1809299 21.137907 148.74245 $t28$ $-$1.888758 21.137907 148.74245 $t29$ $-$ 21.137907 148.74245 (t) 1.5962096 14.211079

## 6 Results and Discussion

From the Table 4 it is observed from all the estimators that,

6.1 Biasednesses are ranging from $-$2.3629 to 4.1947.

6.2 Mean Square Errors are ranging from 14.2111 to 83.5875.

6.3 The PREs of the suggested estimator over estimators in competition are ranging from 112.0830 to 588.1861.

6.4 The sample mean ($𝒕𝟎$), Usual Regression Estimator (Watson,1937) $𝒚¯𝒍⁢𝒓$, Goodman and Hartely’s revised estimator (1958), $𝒕𝟐′$; are unbiased for $𝒀¯$.

6.5 Among the existing estimators, the MSE of the estimator of Yadav and Mishra(2015), $𝒕𝟐𝟑$ is minimum i.e. 15.9282 and MSE of the estimator of Kadilar and Kingi (2003), $𝒕𝟗$ is maximum i.e. 83.5875.

6.6 The suggested estimator is 1.12$%$ more efficient than $𝒕𝟐𝟑$ and 5.88$%$ more efficient than $𝒕𝟗$.

6.7 Since Efficiency is stronger property than the unbiasedness. Hence here we prefer the biased estimator with minimum MSE instead of unbiased estimator with higher MSE.

6.8 It is observed from data that the given below inequalities hold good;

6.9 $𝑴⁢𝑺⁢𝑬⁢(𝒕)≤𝑴⁢𝑺⁢𝑬⁢(𝒕𝟐𝟑)≤𝑽⁢(𝒚¯𝒍⁢𝒓)≤𝑴⁢𝑺⁢𝑬⁢(𝒕𝟏)≤𝑽⁢(𝒚¯)$

6.10 The proposed estimator has minimum MSE and comes out to be more efficient than the other estimators which was our aim of study.

## 7 Conclusion

7.1 We have applied our proposed estimator successfully for the estimation of Average Paddy production. We can also use it for other agricultural areas and productions with larger sample size and population size.

7.2 In case the exact values are not known but we may know the ranges within which our values may be supposed to lie, this Median based estimator will give a precise result.

7.3 The use of this estimator can be extended to other fields also and can be used as an alternative of SRSWOR sample mean when exact values of characteristic under study are not known.

7.4 Whenever taking the auxiliary information involves very high cost we can use the proposed estimator as an alternative of Ratio and Regression type estimators.

7.5 It has been shown theoretically as well as numerically that the suggested estimator is better than the competing estimators for the estimating $Y¯$.

7.6 Mostly the minimum MSE of any ratio type estimator is equal to the variance of the regression estimator but the suggested ratio type estimator has its MSE less than the variance of the usual regression estimator.

7.7 Therefore, the suggested estimator is highly recommended for the practical applications.

## Acknowledgment

The authors are very much thankful to the Editior-in-Chief of JRSS and the learned referees for critically examining the paper which improved the quality of the earlier draft.

## References

Al-Omari, A. I. (2012). Ratio estimation of the population mean using auxiliary information in simple random sampling and median ranked set sampling, Statistics and Probability Letters, 82, p. 1883–18890.

Bahl, S. and Tuteja, R.K. (1991). Ratio and product type exponential estimator, Information and Optimization Sciences, XII (I), p. 159–163.

Chakrabarty, R.P. (1979). Some ratio estimators, Journal of the Indian Society of Agricultural Statistics, 31(1), p. 49–57.

Cochran, W. G. (1940). The estimation of the yields of the cereal experiments by sampling for the ratio of grain to total produce, The Journal of Agric. Science, 30, p. 262–275.

Goodman, L. A. and Hartley, H.O. (1958).The precision of unbiased ratio-type estimators, Journal of the American Statistical Association, 53(282), p. 491–508.

Hartley, H.O. and Ross, A. (1954). Unbiased ratio estimators, Nature, August, 7, Vol. 174, p. 270–271.

Ijaz, M., and Ali, H. (2018). Some improved ratio estimators for estimating mean of finite population, Research and Reviews: Journal of Statistics and Mathematical Sciences, 4(2), 18–23.

Jerajuddin, M. and Kishun, J. (2016). Modified ratio estimators for population mean using size of the sample selected from population, IJSRSET, 2(2), p. 10–16.

Kadilar, C., and Cingi, H. (2003). A study on the chain ratio type estimator, Hacettepe Journal of Mathematics and Statistics, Vol. 32, p. 105–108.

Pandey, H., Yadav, S.K. and Shukla, A.K. (2011). An Improved General Class of Estimators, Estimating Population Mean using Auxiliary Information, International Journal of Statistics and Systems, Vol. 6(1), p. 1–7.

Quenouille, M. H. (1956). Notes on bias in estimation, Biometrika, 43, p. 353–60.

Sahai, A. and Ray, S.K. (1980). Efficient families of ratio and product type estimators, Biometrika, 67(1), p. 211–215.

Singh, G.N. (2003). On the improvement of product method of estimation in sample surveys. Jour. Ind. Soc. Agri. Statistics, 56(3), p. 267–275.

Singh, H. P., Tailor, R., Tailor, R. and Kakran, M. S. (2004): An improved estimator of population mean using power transformation, Journal of the Indian Society of Agricultural Statistics, 58(2), p. 223–230.

Singh, H.P. and Tailor, R. (2003). Use of known correlation coefficient in estimating the finite population mean, Statistics in Transition, 6(4), p. 555–560.

Sisodia, B.V.S. and Dwivedi, V.K. (1981). A modified ratio estimator using coefficient of variation of auxiliary variable, Journal of the Indian Society of Agricultural Statistics, 33, p. 13–18.

Soponviwatkal, K., and Lawson, N. (2017). New ratio estimators for estimating population mean in simple random sampling using a coefficient of variation, correlation coefficient and a regression coefficient, Gazi University Journal of Science, 30(4), p. 610–621.

Subramani, J. and Kumarapandiyan, G. (2012a). Modified ratio estimator for population mean using median of the auxiliary variable, Proceedings of National Conference on Recent developments in the applications of Reliability Theory and Survival Analysis “Application of Reliability Theory and Survival Analysis”, Bonfring Publisher, p. 224–231, ISBN:978-93-82338-25-3.

Subramani, J. and Kumarapandiyan, G. (2012b). Estimation of population mean using co-efficient of variation and median of an auxiliary variable. International Journal of Probability and Statistics, 1(4), p. 111–118.

Subramani, J. and Kumarapandiyan, G. (2012c): Modified ratio estimators for population mean using function of quartiles of auxiliary variable, Bonfring International Journal of Industrial Engineering and Management Science, 2(2), p. 19–23.

Sukhatme, P. V. and Sukhatme, B. V. (1970). Sample Theory of Surveys with Applications, Ames : Iowa State University Press.

Swain, A.K.P.C. (2014). On an improved ratio type estimator of finite population mean in sample surveys, Revista Investigacion Operacional, 35(1), p. 49–57.

Upadhyaya, L. N. and Singh, H. P. (1999). Use of transformed auxiliary variable in estimating the finite population mean, Biometrical Journal, 41(5), p. 627–636.

Watson, D.J. (1937). The estimation of leaf area in field crops, Jour. Agr. Sci., 27(3), p. 474–483.

Yadav, S.K. and Mishra, S.S. (2015). developing improved predictive estimator for finite population mean using auxiliary information, Statistika, 95(1), p. 76–85.

Yadav, S.K., Dixit, M. K., Dungana, H.N. and Mishra, S.S. (2019): Improved estimators for estimating average yield using auxiliary variable, International Journal of Mathematical, Engineering and Management Sciences, Vol. 4, No. 5, p. 1228–1238.

Yadav, S.K, Misra, S., Tiwari, V. and Shukla, A.K. (2010). Improved family of ratio-cum-dual to ratio estimators of finite population mean, IFRSA’s Int. Journal of Computing, 1(1), 17–23.

Yadav, S.K, Misra, S., Tiwari, V. and Shukla, A.K. (2014). Use of coefficient of skewness and quartile deviation of auxiliary variable for improved estimation of population mean, International Journal of Engineering Sciences and Research Technology, 3(1), p. 404–410.

Yan, Z. and Tian, B. (2010). Ratio method to the mean estimation using coefficient of skewness of auxiliary variable, ICICA, Part II, CCIS, 106, p. 103–110.

## Biographies

S. K. Yadav is a faculty in the Department of Statistics at Babasaheb Bhimrao Ambedkar University Lucknow, U.P., India. He earned his M.Sc. and Ph.D. degrees in Statistics from Lucknow University and qualified the National Eligibility Test. Dr. Yadav has published more than 40 research papers in national and international journals of repute indexed in Scopus/Web of science and two books from an international publisher. He is a referee for 20 reputed international journals. He has been awarded Young Scientist Award in 2016 for the contribution in the field survey sampling by Venus International Research Foundation, Chennai, India and best paper award in 2018 by MTMI, USA. He has presented papers in more than 20 national and international conferences and also delivered 18 invited talks in several conferences.

Shanya Baghel is pursuing her Ph.D in IIT(ISM) Dhanbad, Department of Mathematics and Computing as a JRF. She has completed her M.Phil. from Babasaheb Bhimrao Ambedkar University (BBAU), Lucknow, in 2020. She received her M.Sc. degree in Applied Statistics from BBAU, Lucknow in 2018. She attended University of Lucknow, India and received her B.Sc. in 2016. Her M.Phil work centers on Estimation of Population Mean using known Auxiliary Variable.

Sugandha Saxena is a postgraduate from Babasaheb Bhimrao Ambedkar University. She is an alumna of City Montessori School, Lucknow, India. She received a B.Sc. degree in Statistics and Mathematics from Mahila P.G. College, University of Lucknow. She completed her M.Sc. in Applied Statistics from Babasaheb Bhimrao Ambedkar University, Lucknow.

Avinash Kumar Singh received his M.Sc. degree in Applied Statistics form Babasaheb Bhimrao Ambedkar University, Lucknow, India. He attended the University of Lucknow, India and received his B.Sc. in Physics, Mathematics and Statistics.