3. Name:SumitSanjaySatam,(F.Y.:S-1)
20) Person’s coeff of kurtosis
β2 = μ4 / (μ2)2, γ2 = β2 -3
21) Covariance=cov(X,Y)= ∑(X - X¯) (Y - Y¯) / n
= ∑XY / n - X¯ Y¯
22) Karl Pearson’s coeff of correlation
r = cov(X,Y) / σx σy (-1<= r <=1)
23) Coeff of determination r2 = Explained var. / Total var.
24) Person’s rank correln. Coeff R= 1 – ( 6∑(d i)2 / n(n2-1))
25) Regression lines : y on x
Y = a + bX / y - y¯ = byx (x - x¯)
byx = r (σy / σx) = cov(x,y) / (σ x) 2
= (∑xy - nX¯ Y¯) / (∑x2 – n(x¯)2
)
26) Regression lines : x on y
X = a + bY / x - x¯ = bxy (y - y¯)
byx = r (σx / σy) = cov(x,y) / (σ y) 2
= (∑xy - nX¯ Y¯) / (∑y2 – n (y¯)2
)
27) Least square fit y = a + bx
∑y = na + b∑x
∑xy = a ∑x + b ∑x2
28) Coeff of correlation r = √(bxy * byx)
4. Name:SumitSanjaySatam,(F.Y.:S-1)
29) Fitting second degree curve
Y = a + bX + cX2
∑y = an + b∑x + c∑x2
∑xy = a∑x + b∑x2 + c∑x3
∑x2y = a∑x2 + b∑x3 + c∑x4
30) Fitting of exponential curve y = abx
Log y = log a + x log b
V = A + xB
∑V = nA + B∑x
∑Vx = A∑x + B∑x2
31) Eqn of plane of regression of X1 on X2 & X3
X1 = a + b12.3 X2 + b13.2 X3
X2 = a + b21.3 X1 + b23.1 X3
X3 = a + b31.2 X1 + b32.1 X3
32) Partial regression coeff of eqn of X1 on X2 & X3
b12.3 = (σ1 /σ2) * (r12 - r13 r23 / 1- (r23)2
) = -(σ1 / σ2) (R12 / R11)
b13.2 = (σ1 /σ3) * (r13 - r12 r23 / 1- (r23)2
) = -(σ1 / σ3) (R13 / R11)
R11, R12, R13 are cofactors in matrix R
33) Multiple regression eqn, deviations taken from means Xi=Xi –(Xi)¯
Regression eqn of X1 on X2 & X3
(R11 / σ1) * x1 + (R12 / σ2) * x2 + (R13 / σ3) * x3 = 0
(R21 / σ1) * x1 + (R22 / σ2) * x2 + (R23 / σ3) * x3 = 0
(R31 / σ1) * x1 + (R32 / σ2) * x2 + (R33 / σ3) * x3 = 0
5. Name:SumitSanjaySatam,(F.Y.:S-1)
34) Multiple correlation coeff
R1.23 = √((r12)2
+(r13)2
- 2r12 r13 r23 ) / (1- (r23)2
)
R2.13 = √((r12)2
+(r23)2
- 2r12 r13 r23 ) / (1- (r23)2
)
R3.12 = √((r13)2
+(r23)2
- 2r12 r13 r23 ) / (1- (r12)2
)
(0<= Ri.jk <= 1)
35) Partial correlation coeff
r12.3 = (r12 - r13 r23 ) / √((1- (r13)2
) (1- (r23)2
)
r13.2 = (r13 - r12 r23 ) / √((1- (r12)2
) (1- (r23)2
)
r23.1 = (r23 - r12 r13 ) / √((1- (r12)2
) (1- (r13)2
)
(-1<= ri.jk <= 1)
35) m –yearly moving average for the time series
T t1 t2 t3 -------- tn
Y y1 y2 y3 -------- yn
y1¯ = (y1 + y2 + -------- +ym)/m , y2¯ = (y2 + y3 + -------- +ym+1)/m
(centering of moving average)
36) Fitting linear tend y = a + bt
Use ∑y = na + b∑X
∑yX = a∑X + b∑x2 , X = t - A, A = mid pt of time
= (½) (2middle year)
37) Exponential smoothing
S0 = y0
S1 = α yt + (1- α) St-1 0< α <1
38) Methods of measuring seasonal variations
a) Ratio to trend method
6. Name:SumitSanjaySatam,(F.Y.:S-1)
(y /T)*100
Correction factor= ((1200/Total of indices) / (400/Total of indices)
b) Link relative method
Link relative for any period
= (current period figure / previous period figure) *100
Chain relative
= (C.R. of previous period * L.R. of current period) /100
Thanks for Reading