Subject: Statistics
Paper: Multivariate Analysis
Module: Canonical Correlation
Development Team
Principal investigator: Dr. Bhaswati Ganguli, Professor, Department of Statistics, University of Calcutta
Paper co-ordinator: Dr. Sugata SenRoy,Professor, Department of Statistics, University of Calcutta
Content writer: Souvik Bandyopadhyay, Senior Lecturer, Indian Institute of Public Health, Hyderabad
Content reviewer: Dr. Kalyan Das,Professor, Department of Statistics, University of Calcutta
Development Team
Principal investigator: Dr. Bhaswati Ganguli, Professor, Department of Statistics, University of Calcutta
Paper co-ordinator: Dr. Sugata SenRoy,Professor, Department of Statistics, University of Calcutta
Content writer: Souvik Bandyopadhyay, Senior Lecturer, Indian Institute of Public Health, Hyderabad
Content reviewer: Dr. Kalyan Das,Professor, Department of Statistics, University of Calcutta
Development Team
Principal investigator: Dr. Bhaswati Ganguli, Professor, Department of Statistics, University of Calcutta
Paper co-ordinator: Dr. Sugata SenRoy,Professor, Department of Statistics, University of Calcutta
Content writer: Souvik Bandyopadhyay, Senior Lecturer, Indian Institute of Public Health, Hyderabad
Content reviewer: Dr. Kalyan Das,Professor, Department of Statistics, University of Calcutta
Development Team
Principal investigator: Dr. Bhaswati Ganguli, Professor, Department of Statistics, University of Calcutta
Paper co-ordinator: Dr. Sugata SenRoy,Professor, Department of Statistics, University of Calcutta
Content writer: Souvik Bandyopadhyay, Senior Lecturer, Indian Institute of Public Health, Hyderabad
Content reviewer: Dr. Kalyan Das,Professor, Department of Statistics, University of Calcutta
Correlations
I Q : What is canonical correlation ?
I It’s a technique which seeks to identify and quantify the relationship between two sets of variables.
I Special cases
I Simple correlation
I Multiple Correlation
I In simple correlation, we study the relationship between two variables X andY.
I In multiple correlation, we study the relationship between a variable Y and a set of variables (X1, . . . , Xp).
←To do this we take a linear combination ofX1, . . . , Xp and find its correlation with Y.
Correlations
I Q : What is canonical correlation ?
I It’s a technique which seeks to identify and quantify the relationship between two sets of variables.
I Special cases
I Simple correlation
I Multiple Correlation
I In simple correlation, we study the relationship between two variables X andY.
I In multiple correlation, we study the relationship between a variable Y and a set of variables (X1, . . . , Xp).
←To do this we take a linear combination ofX1, . . . , Xp and find its correlation with Y.
Correlations
I Q : What is canonical correlation ?
I It’s a technique which seeks to identify and quantify the relationship between two sets of variables.
I Special cases
I Simple correlation
I Multiple Correlation
I In simple correlation, we study the relationship between two variables X andY.
I In multiple correlation, we study the relationship between a variable Y and a set of variables (X1, . . . , Xp).
←To do this we take a linear combination ofX1, . . . , Xp and find its correlation with Y.
Correlations
I Q : What is canonical correlation ?
I It’s a technique which seeks to identify and quantify the relationship between two sets of variables.
I Special cases
I Simple correlation
I Multiple Correlation
I In simple correlation, we study the relationship between two variables X andY.
I In multiple correlation, we study the relationship between a variable Y and a set of variables (X1, . . . , Xp).
←To do this we take a linear combination ofX1, . . . , Xp and find its correlation with Y.
Correlations
I Q : What is canonical correlation ?
I It’s a technique which seeks to identify and quantify the relationship between two sets of variables.
I Special cases
I Simple correlation
I Multiple Correlation
I In simple correlation, we study the relationship between two variables X andY.
I In multiple correlation, we study the relationship between a variable Y and a set of variables (X1, . . . , Xp).
←To do this we take a linear combination ofX1, . . . , Xp and find its correlation with Y.
Correlations
I Q : What is canonical correlation ?
I It’s a technique which seeks to identify and quantify the relationship between two sets of variables.
I Special cases
I Simple correlation
I Multiple Correlation
I In simple correlation, we study the relationship between two variables X andY.
I In multiple correlation, we study the relationship between a variable Y and a set of variables (X1, . . . , Xp).
←To do this we take a linear combination ofX1, . . . , Xp and find its correlation with Y.
Correlations
I Q : What is canonical correlation ?
I It’s a technique which seeks to identify and quantify the relationship between two sets of variables.
I Special cases
I Simple correlation
I Multiple Correlation
I In simple correlation, we study the relationship between two variables X andY.
I In multiple correlation, we study the relationship between a variable Y and a set of variables (X1, . . . , Xp).
←To do this we take a linear combination ofX1, . . . , Xp and find its correlation with Y.
Canonical Correlation
I In canonical correlation we study the relationship between two sets of variables (X1, . . . , Xr)and (X1∗, . . . , Xs∗).
I Canonical correlation requires that each set of variables be reduced to a single variable and then finding their correlation.
I Usually these two variables are found by taking linear combinations of the variables in each set, under certain pre-fixed criteria.
I The variables obtained by these linear combinations are known as canonical variables and the correlation between them as canonical correlation.
Canonical Correlation
I In canonical correlation we study the relationship between two sets of variables (X1, . . . , Xr)and (X1∗, . . . , Xs∗).
I Canonical correlation requires that each set of variables be reduced to a single variable and then finding their correlation.
I Usually these two variables are found by taking linear combinations of the variables in each set, under certain pre-fixed criteria.
I The variables obtained by these linear combinations are known as canonical variables and the correlation between them as canonical correlation.
Canonical Correlation
I In canonical correlation we study the relationship between two sets of variables (X1, . . . , Xr)and (X1∗, . . . , Xs∗).
I Canonical correlation requires that each set of variables be reduced to a single variable and then finding their correlation.
I Usually these two variables are found by taking linear combinations of the variables in each set, under certain pre-fixed criteria.
I The variables obtained by these linear combinations are known as canonical variables and the correlation between them as canonical correlation.
Canonical Correlation
I In canonical correlation we study the relationship between two sets of variables (X1, . . . , Xr)and (X1∗, . . . , Xs∗).
I Canonical correlation requires that each set of variables be reduced to a single variable and then finding their correlation.
I Usually these two variables are found by taking linear combinations of the variables in each set, under certain pre-fixed criteria.
I The variables obtained by these linear combinations are known as canonical variables and the correlation between them as canonical correlation.
The Model
I Let there ber variables in the1st groupX1= (X1, . . . , Xr) ands variables in the2nd groupX2 = (Xr+1, . . . , Xr+s).
I Assume, without loss of generality, thatr ≤s, i.e. take the smaller group of variables as the 1st group.
I Also let
E(X1) =µ1 andE(X2) =µ2 V ar(X1) = Σ11,V ar(X2) = Σ22
andCov(X1,X2) = Σ12 (orΣ021)
I DefineX= X1
X2
,µ=
µ1 µ2
andΣ =
Σ11 Σ12 Σ21 Σ22
.
The Model
I Let there ber variables in the1st groupX1= (X1, . . . , Xr) ands variables in the2nd groupX2 = (Xr+1, . . . , Xr+s).
I Assume, without loss of generality, thatr ≤s, i.e. take the smaller group of variables as the 1st group.
I Also let
E(X1) =µ1 andE(X2) =µ2 V ar(X1) = Σ11,V ar(X2) = Σ22
andCov(X1,X2) = Σ12 (orΣ021)
I DefineX= X1
X2
,µ=
µ1 µ2
andΣ =
Σ11 Σ12 Σ21 Σ22
.
The Model
I Let there ber variables in the1st groupX1= (X1, . . . , Xr) ands variables in the2nd groupX2 = (Xr+1, . . . , Xr+s).
I Assume, without loss of generality, thatr ≤s, i.e. take the smaller group of variables as the 1st group.
I Also let
E(X1) =µ1 andE(X2) =µ2 V ar(X1) = Σ11,V ar(X2) = Σ22
andCov(X1,X2) = Σ12 (orΣ021)
I DefineX= X1
X2
,µ=
µ1 µ2
andΣ =
Σ11 Σ12 Σ21 Σ22
.
The Model
I Let there ber variables in the1st groupX1= (X1, . . . , Xr) ands variables in the2nd groupX2 = (Xr+1, . . . , Xr+s).
I Assume, without loss of generality, thatr ≤s, i.e. take the smaller group of variables as the 1st group.
I Also let
E(X1) =µ1 andE(X2) =µ2 V ar(X1) = Σ11,V ar(X2) = Σ22
andCov(X1,X2) = Σ12 (orΣ021)
I DefineX= X1
X2
,µ=
µ1 µ2
andΣ =
Σ11 Σ12 Σ21 Σ22
.
The Problem
I Σ12 containsrselements which gives the correlations between each variable of the 1st group with those of the2nd group
← gives a perfect picture of the relationships
← however, not easily comprehendible.
I Instead, can we look at just a few values, which would make understanding easier ?
The Problem
I Σ12 containsrselements which gives the correlations between each variable of the 1st group with those of the2nd group
← gives a perfect picture of the relationships
← however, not easily comprehendible.
I Instead, can we look at just a few values, which would make understanding easier ?
The Method
I Let aand bbe respectively r andsdimensional vectors.
I Define the variables,
U =a0X1 and V =b0X2.
Then E(U) =a0µ1, E(V) =b0µ2, V ar(U) =a0Σ11a, V ar(V) =b0Σ22b, with Cov(U, V) =a0Σ12b.
so that
Corr(U, V) = a0Σ12b
√a0Σ11a×b0Σ22b.
The Method
I Let aand bbe respectively r andsdimensional vectors.
I Define the variables,
U =a0X1 and V =b0X2.
Then E(U) =a0µ1, E(V) =b0µ2, V ar(U) =a0Σ11a, V ar(V) =b0Σ22b, with Cov(U, V) =a0Σ12b.
so that
Corr(U, V) = a0Σ12b
√a0Σ11a×b0Σ22b.
How to choose a and b
I The coefficients a1,b1 of the first pair(U1, V1)are so chosen as to maximizeCorr(U, V).
I The 2nd pair(U2, V2) are chosen to maximizeCorr(U, V) subject to their combining vectorsa2 andb2 being orthogonal to a1 and b1, respectively.
I In general, the jth pair (Uj, Vj) are chosen to maximize Corr(U, V)subject toaj and bj being orthogonal to ak and bk, respectively, for all k < j.
I This can be done till j=r.
I Usually for all j= 1, . . . , r,
a0jaj = 1 and b0jbj = 1.
How to choose a and b
I The coefficients a1,b1 of the first pair(U1, V1)are so chosen as to maximizeCorr(U, V).
I The 2nd pair(U2, V2) are chosen to maximizeCorr(U, V) subject to their combining vectorsa2 andb2 being orthogonal to a1 and b1, respectively.
I In general, the jth pair (Uj, Vj) are chosen to maximize Corr(U, V)subject toaj and bj being orthogonal to ak and bk, respectively, for all k < j.
I This can be done till j=r.
I Usually for all j= 1, . . . , r,
a0jaj = 1 and b0jbj = 1.
How to choose a and b
I The coefficients a1,b1 of the first pair(U1, V1)are so chosen as to maximizeCorr(U, V).
I The 2nd pair(U2, V2) are chosen to maximizeCorr(U, V) subject to their combining vectorsa2 andb2 being orthogonal to a1 and b1, respectively.
I In general, the jth pair (Uj, Vj) are chosen to maximize Corr(U, V)subject toaj and bj being orthogonal to ak and bk, respectively, for all k < j.
I This can be done till j=r.
I Usually for all j= 1, . . . , r,
a0jaj = 1 and b0jbj = 1.
How to choose a and b
I The coefficients a1,b1 of the first pair(U1, V1)are so chosen as to maximizeCorr(U, V).
I The 2nd pair(U2, V2) are chosen to maximizeCorr(U, V) subject to their combining vectorsa2 andb2 being orthogonal to a1 and b1, respectively.
I In general, the jth pair (Uj, Vj) are chosen to maximize Corr(U, V)subject toaj and bj being orthogonal to ak and bk, respectively, for all k < j.
I This can be done till j=r.
I Usually for all j= 1, . . . , r,
a0jaj = 1 and b0jbj = 1.
How to choose a and b
I The coefficients a1,b1 of the first pair(U1, V1)are so chosen as to maximizeCorr(U, V).
I The 2nd pair(U2, V2) are chosen to maximizeCorr(U, V) subject to their combining vectorsa2 andb2 being orthogonal to a1 and b1, respectively.
I In general, the jth pair (Uj, Vj) are chosen to maximize Corr(U, V)subject toaj and bj being orthogonal to ak and bk, respectively, for all k < j.
I This can be done till j=r.
I Usually for all j= 1, . . . , r,
a0jaj = 1 and b0jbj = 1.
Result
I Let λ1 ≥λ2≥. . .≥λr be the eigenvalues of
Σ−1/211 Σ12Σ−122Σ21Σ−1/211 andp1,p2, . . . ,pr the corresponding eigen-vectors.
I In fact, λ1 ≥λ2≥. . .≥λr are also the r largest (out ofs) eigen-values of Σ−1/222 Σ21Σ−111Σ12Σ−1/222 with q1,q2, . . . ,qr the corresponding eigen-vectors.
Result
Corr(U1, V1) =ρ1=p λ1, witha1= Σ−1/211 p1 andb1= Σ−1/222 q1.
I Thus U1 =p01Σ−1/211 X1 andV2=q01Σ−1/222 X2.
Result
I Let λ1 ≥λ2≥. . .≥λr be the eigenvalues of
Σ−1/211 Σ12Σ−122Σ21Σ−1/211 andp1,p2, . . . ,pr the corresponding eigen-vectors.
I In fact, λ1 ≥λ2≥. . .≥λr are also the r largest (out ofs) eigen-values of Σ−1/222 Σ21Σ−111Σ12Σ−1/222 with q1,q2, . . . ,qr the corresponding eigen-vectors.
Result
Corr(U1, V1) =ρ1=p λ1,
witha1= Σ−1/211 p1 andb1= Σ−1/222 q1.
−1/2 −1/2
Result
I Let λ1 ≥λ2≥. . .≥λr be the eigenvalues of
Σ−1/211 Σ12Σ−122Σ21Σ−1/211 andp1,p2, . . . ,pr the corresponding eigen-vectors.
I In fact, λ1 ≥λ2≥. . .≥λr are also the r largest (out ofs) eigen-values of Σ−1/222 Σ21Σ−111Σ12Σ−1/222 with q1,q2, . . . ,qr the corresponding eigen-vectors.
Result
Corr(U1, V1) =ρ1=p λ1,
witha1= Σ−1/211 p1 andb1= Σ−1/222 q1.
I Thus U1 =p01Σ−1/211 X1 andV2=q01Σ−1/222 X2.
Result
I Let λ1 ≥λ2≥. . .≥λr be the eigenvalues of
Σ−1/211 Σ12Σ−122Σ21Σ−1/211 andp1,p2, . . . ,pr the corresponding eigen-vectors.
I In fact, λ1 ≥λ2≥. . .≥λr are also the r largest (out ofs) eigen-values of Σ−1/222 Σ21Σ−111Σ12Σ−1/222 with q1,q2, . . . ,qr the corresponding eigen-vectors.
Result
Corr(U1, V1) =ρ1=p λ1,
witha1= Σ−1/211 p1 andb1= Σ−1/222 q1.
−1/2 −1/2
Result (contd.)
Result
In general, forj = 1, . . . , r,
Corr(Uj, Vj) =ρj =p λj,
withaj = Σ−1/211 pj andbj = Σ−1/222 qj.
I Usually, only the first one or two of the canonical correlations are looked at.
Result (contd.)
Result
In general, forj = 1, . . . , r,
Corr(Uj, Vj) =ρj =p λj,
withaj = Σ−1/211 pj andbj = Σ−1/222 qj.
I Usually, only the first one or two of the canonical correlations are looked at.
Proof
I Since Σ11and Σ22 are p.d., define
c= Σ1/211 a,d= Σ1/222 b so that a= Σ−1/211 c,b= Σ−1/222 d.
Corr(U, V) = c0Σ−1/211 Σ12Σ−1/222 d q
c0Σ−1/211 Σ11Σ−1/211 c×d0Σ−1/222 Σ22Σ−1/222 d i.e. Corr(U, V) = c0Σ−1/211 Σ12Σ−1/222 d
√
c0c×d0d By Cauchy-Schwarz inequality,
c0Σ−1/211 Σ12Σ−1/222 d≤(c0Σ−1/211 Σ12Σ−122Σ21Σ−1/211 c)1/2(d0d)1/2
Proof (contd.)
I Σ−1/211 Σ12Σ−122Σ21Σ−1/211 is symmetric, with largest eigen-value λ1 and corresponding eigen-vector p1. Thus
c0Σ−1/211 Σ12Σ−122Σ21Σ−1/211 c≤λ1c0c=λ1.
I Equality occurs atc=p1.
I Also equality in the Cauchy-Schwarz inequality occurs if d is proportional toΣ−1/222 Σ21Σ−1/211 p1.
I This means b is proportional to
Σ−1/222 Σ−1/222 Σ21Σ−1/211 p1 = Σ−1/222 q1 (say), where, q = Σ−1/2Σ Σ−1/2p .
Proof (contd.)
I Σ−1/211 Σ12Σ−122Σ21Σ−1/211 is symmetric, with largest eigen-value λ1 and corresponding eigen-vector p1. Thus
c0Σ−1/211 Σ12Σ−122Σ21Σ−1/211 c≤λ1c0c=λ1.
I Equality occurs atc=p1.
I Also equality in the Cauchy-Schwarz inequality occurs if d is proportional toΣ−1/222 Σ21Σ−1/211 p1.
I This means b is proportional to
Σ−1/222 Σ−1/222 Σ21Σ−1/211 p1 = Σ−1/222 q1 (say), where, q1 = Σ−1/222 Σ21Σ−1/211 p1.
Proof (contd.)
I Σ−1/211 Σ12Σ−122Σ21Σ−1/211 is symmetric, with largest eigen-value λ1 and corresponding eigen-vector p1. Thus
c0Σ−1/211 Σ12Σ−122Σ21Σ−1/211 c≤λ1c0c=λ1.
I Equality occurs atc=p1.
I Also equality in the Cauchy-Schwarz inequality occurs if d is proportional toΣ−1/222 Σ21Σ−1/211 p1.
I This means b is proportional to
Σ−1/222 Σ−1/222 Σ21Σ−1/211 p1 = Σ−1/222 q1 (say), where, q = Σ−1/2Σ Σ−1/2p .
Proof (contd.)
I Σ−1/211 Σ12Σ−122Σ21Σ−1/211 is symmetric, with largest eigen-value λ1 and corresponding eigen-vector p1. Thus
c0Σ−1/211 Σ12Σ−122Σ21Σ−1/211 c≤λ1c0c=λ1.
I Equality occurs atc=p1.
I Also equality in the Cauchy-Schwarz inequality occurs if d is proportional toΣ−1/222 Σ21Σ−1/211 p1.
I This means b is proportional to
Σ−1/222 Σ−1/222 Σ21Σ−1/211 p1 = Σ−1/222 q1 (say), where, q1 = Σ−1/222 Σ21Σ−1/211 p1.
Proof (contd.)
I Notice that from properties of eigen-values, Σ−1/211 Σ12Σ−122Σ21Σ−1/211 p1 =λ1p1
so that premultiplying both sides byΣ−1/222 Σ21Σ−1/211 yields for q1 = Σ−1/222 Σ21Σ−1/211 p1
Σ−1/222 Σ21Σ−111Σ12Σ−1/222 q1 =λ1q1
i.e. λ1 is an eigen-value ofΣ−1/222 Σ21Σ−111Σ12Σ−1/222 with corresponding eigen-vectorq1.
I Thus with a1 = Σ−1/211 p1 and b1 = Σ−1/222 q1,
0 0 p
Proof (contd.)
I Notice that from properties of eigen-values, Σ−1/211 Σ12Σ−122Σ21Σ−1/211 p1 =λ1p1
so that premultiplying both sides byΣ−1/222 Σ21Σ−1/211 yields for q1 = Σ−1/222 Σ21Σ−1/211 p1
Σ−1/222 Σ21Σ−111Σ12Σ−1/222 q1 =λ1q1
i.e. λ1 is an eigen-value ofΣ−1/222 Σ21Σ−111Σ12Σ−1/222 with corresponding eigen-vectorq1.
I Thus with a1 = Σ−1/211 p1 and b1 = Σ−1/222 q1,
Corr(U1, V1) =maxa,bCorr(a0X1,b0X2) =p λ1.
Properties
I E(Uj) =p0jΣ−1/211 µ1 andE(Vj) =q0jΣ−1/222 µ2 for all j.
I V ar(Uj) =p0jΣ−1/211 Σ11Σ−1/211 pj =p0jpj = 1 for all j.
I V ar(Vj) =q0jΣ−1/222 Σ22Σ−1/222 qj =q0jqj = 1 for all j.
I Cov(Uj, Uk) =p0jΣ−1/211 Σ11Σ−1/211 pk=p0jpk= 0 for all k6=j.
I Cov(Vj, Vk) =q0jΣ−1/222 Σ22Σ−1/222 qk =q0jqk = 0for all k6=j.
I Cov(Uj, Vk) =p0jΣ−1/211 Σ12Σ−1/222 qk=p0jpk= 0 for all j6=k.
Properties
I E(Uj) =p0jΣ−1/211 µ1 andE(Vj) =q0jΣ−1/222 µ2 for all j.
I V ar(Uj) =p0jΣ−1/211 Σ11Σ−1/211 pj =p0jpj = 1 for all j.
I V ar(Vj) =q0jΣ−1/222 Σ22Σ−1/222 qj =q0jqj = 1 for all j.
I Cov(Uj, Uk) =p0jΣ−1/211 Σ11Σ−1/211 pk=p0jpk= 0 for all k6=j.
I Cov(Vj, Vk) =q0jΣ−1/222 Σ22Σ−1/222 qk =q0jqk = 0for all k6=j.
I Cov(Uj, Vk) =p0jΣ−1/211 Σ12Σ−1/222 qk=p0jpk= 0 for all j6=k.
Properties
I E(Uj) =p0jΣ−1/211 µ1 andE(Vj) =q0jΣ−1/222 µ2 for all j.
I V ar(Uj) =p0jΣ−1/211 Σ11Σ−1/211 pj =p0jpj = 1 for all j.
I V ar(Vj) =q0jΣ−1/222 Σ22Σ−1/222 qj =q0jqj = 1 for all j.
I Cov(Uj, Uk) =p0jΣ−1/211 Σ11Σ−1/211 pk=p0jpk= 0 for all k6=j.
I Cov(Vj, Vk) =q0jΣ−1/222 Σ22Σ−1/222 qk =q0jqk = 0for all k6=j.
I Cov(Uj, Vk) =p0jΣ−1/211 Σ12Σ−1/222 qk=p0jpk= 0 for all j6=k.
Properties
I E(Uj) =p0jΣ−1/211 µ1 andE(Vj) =q0jΣ−1/222 µ2 for all j.
I V ar(Uj) =p0jΣ−1/211 Σ11Σ−1/211 pj =p0jpj = 1 for all j.
I V ar(Vj) =q0jΣ−1/222 Σ22Σ−1/222 qj =q0jqj = 1 for all j.
I Cov(Uj, Uk) =p0jΣ−1/211 Σ11Σ−1/211 pk=p0jpk= 0 for all k6=j.
I Cov(Vj, Vk) =q0jΣ−1/222 Σ22Σ−1/222 qk =q0jqk = 0for all k6=j.
I Cov(Uj, Vk) =p0jΣ−1/211 Σ12Σ−1/222 qk=p0jpk= 0 for all j6=k.
Properties
I E(Uj) =p0jΣ−1/211 µ1 andE(Vj) =q0jΣ−1/222 µ2 for all j.
I V ar(Uj) =p0jΣ−1/211 Σ11Σ−1/211 pj =p0jpj = 1 for all j.
I V ar(Vj) =q0jΣ−1/222 Σ22Σ−1/222 qj =q0jqj = 1 for all j.
I Cov(Uj, Uk) =p0jΣ−1/211 Σ11Σ−1/211 pk=p0jpk= 0 for all k6=j.
I Cov(Vj, Vk) =q0jΣ−1/222 Σ22Σ−1/222 qk =q0jqk = 0for all k6=j.
I Cov(Uj, Vk) =p0jΣ−1/211 Σ12Σ−1/222 qk=p0jpk= 0 for all j6=k.
Properties
I E(Uj) =p0jΣ−1/211 µ1 andE(Vj) =q0jΣ−1/222 µ2 for all j.
I V ar(Uj) =p0jΣ−1/211 Σ11Σ−1/211 pj =p0jpj = 1 for all j.
I V ar(Vj) =q0jΣ−1/222 Σ22Σ−1/222 qj =q0jqj = 1 for all j.
I Cov(Uj, Uk) =p0jΣ−1/211 Σ11Σ−1/211 pk=p0jpk= 0 for all k6=j.
I Cov(Vj, Vk) =q0jΣ−1/222 Σ22Σ−1/222 qk =q0jqk = 0for all k6=j.
I Cov(Uj, Vk) =p0jΣ−1/211 Σ12Σ−1/222 qk=p0jpk= 0 for all j6=k.
Summary
I The idea of canonical correlation is discussed.
I The expressions for the canonical correlations are derived.
I Some properties of canonical variables are discussed.
Summary
I The idea of canonical correlation is discussed.
I The expressions for the canonical correlations are derived.
I Some properties of canonical variables are discussed.
Summary
I The idea of canonical correlation is discussed.
I The expressions for the canonical correlations are derived.
I Some properties of canonical variables are discussed.