@

Errata for "Introduction to Nonparametric Regression"

(25) Page 124

Wrong: In the description of AIC, the variance of error of the normal distribution is not included in the parameters.

Right: When the normal distribution is assumed, the number of parameters in AIC is the trace of Hat matrix plus one.

(24) Page 236, Eq.(4.12)

Wrong: t>0, t=0

Right: t_i>0, t_i=0

(23) Page 159, Eq.(3.178)

Wrong: X_N

Right: X_n

(22) Page 527, 13th and 14th lines of (4)

Wrong: xa2 <- seq(from = ex[1], to = ex[length(ex)], length = 4)

Right: xa2 <- seq(from = ex1[1], to = ex1[length(ex1)], length = 4)

(21) Page 128, the last line

Wrong: (Chapter 5 of [2])

Right: this part should be deleted.

(20) Page 225, page 405.

Wrong: K.P. Burnhan

Right: K.P. Burnham

(19) Page 513, first and second lines of (5).

Wrong: g.cov <- function(x1, x2, theta){exp(-(rdist(x1, x2)/theta)**2)}

Right: g.cov <<- function(x1, x2, theta){exp(-(rdist(x1, x2)/theta)**2)}

(18) Page 511, first and second lines of (5).

Wrong: g.cov <- function(x1, x2, theta){exp( - (rdist(x1, x2) / theta) **2)}

Right: g.cov <<- function(x1, x2, theta){exp( - (rdist(x1, x2) / theta) **2)}

(17) Internal knots of Figure 2.19, Figure 2.20, Figure 2.21, and Figure 2.22 are located at

Wrong: {4, 7, 11, 15, 19, 23, 27}.

Right: {4, 7, 10, 13, 16, 19, 22, 25, 28}.

(16) Page 306, first line of (K) object, page 308, first line of (3), and page 514, fist line of (K) object

Wrong: (nd, ne, xx1, xx2, yy, ex1, ex2, ey1)

Right: (nd, ne, xx1, xx2, yy, ex1, ex2)

(15) Page 199, (2) and (4) of (B).

Wrong:

(2) fit.ks <- ksmooth(x1, y1, "normal", bandwidth = bwsplus)

(4) fit.ks <- ksmooth(x2, y2, "normal", bandwidth = bwsplus)

Right:

(2) fit.ks <- ksmooth(x1, y1, "normal", bandwidth = bwsplus, x.points = x1)

(4) fit.ks <- ksmooth(x2, y2, "normal", bandwidth = bwsplus, x.points = x2)

(14) Page 502, (E) of B.2.

Wrong:

The S-Plus object, in which the part of (2) is modified as below, is used as an R object.

# (2)

data1 <- data.frame(x = xx, y = yy)

Right:

The S-Plus object, in which the parts of (2) and (4) are modified as below, is used as an R object.

# (2)

data1 <- data.frame(x = xx, y = yy)

# (4)

fit.lm <- lm(y ~ poly(x, degree = 11), data = data1)

ey <- predict(fit.lm, newdata = data2)

(13) Page 79, (4) of (A).

Wrong:

for(ii in 1.:nd) {

ey <- mean(y2[ii:(ii + 2. * mm)])

}

Right:

for(ii in 1.:nd) {

ey[ii] <- mean(y2[ii:(ii + 2. * mm)])

}

(12) Page 440, Line 12.

Wrong: "when $\{ {\bf z}_{i}^{(g)} \}$ are used as feature vectors,"

Right: "when $\{ z_{i}^{(g)} \}$ are used as feature vectors,"

(11) Page 416, eq(7.20).

Wrong: "{\hat {\bf V}^{(g)}} = \frac{1}{n_g -1}\sum_{i=1}^{n_g}\sum_{l=1}^{n_g}( {\bf x}_{i}^{(g)} - \mbox{\boldmath$\hat \mu$}^{(g)}) ({\bf x}_{l}^{(g)} - \mbox{\boldmath$\mu$}^{(g)})^t"

Right: "{\hat {\bf V}^{(g)}} = \frac{1}{n_g -1}\sum_{i=1}^{n_g}( {\bf x}_{i}^{(g)} - \mbox{\boldmath$\hat \mu$}^{(g)}) ({\bf x}_{i}^{(g)} - \mbox{\boldmath$\mu$}^{(g)})^t"

(11) Page 416, eq(7.20).

Wrong: "{\hat {\bf V}^{(g)}} = \frac{1}{n_g -1}\sum_{i=1}^{n_g}\sum_{l=1}^{n_g}( {\bf x}_{i}^{(g)} - \mbox{\boldmath$\hat \mu$}^{(g)}) ({\bf x}_{l}^{(g)} - \mbox{\boldmath$\mu$}^{(g)})^t"

Right: "{\hat {\bf V}^{(g)}} = \frac{1}{n_g -1}\sum_{i=1}^{n_g}( {\bf x}_{i}^{(g)} - \mbox{\boldmath$\hat \mu$}^{(g)}) ({\bf x}_{i}^{(g)} - \mbox{\boldmath$\mu$}^{(g)})^t"

(10) Page 416, eq(7.19).

Wrong: "{\hat {\bf V}^{(g)}} = \frac{1}{n_g}\sum_{i=1}^{n_g}\sum_{l=1}^{n_g}( {\bf x}_{i}^{(g)} - \mbox{\boldmath$\hat \mu$}^{(g)}) ({\bf x}_{l}^{(g)} - \mbox{\boldmath$\hat \mu$}^{(g)})^t"

Right: "{\hat {\bf V}^{(g)}} = \frac{1}{n_g}\sum_{i=1}^{n_g}( {\bf x}_{i}^{(g)} - \mbox{\boldmath$\hat \mu$}^{(g)}) ({\bf x}_{i}^{(g)} - \mbox{\boldmath$\hat \mu$}^{(g)})^t"

(9) Page 416, Line 7 and 8.

Wrong: "a feature vector of a data point which is a member of $C_j$"

Right: "a feature vector of a data point which is a member of $C_g$"

(8) Page 413, eq(7.12).

Wrong: "\sum_{g=1}^G n_j - \sum_{g=1}^G \int_{\Omega_j} n_j p_j ({\bf x} ){\bf dx}"

Right: "\sum_{g=1}^G n_g - \sum_{g=1}^G \int_{\Omega_g} n_g p_g ({\bf x} ){\bf dx}"

(7) Page 90, (1) Line 4 and 5.

Wrong: "is a bandwidth (\verb+h+ in"

Right: "is a bandwidth ($h$ in"

(6) Page 394, Line 3.

Wrong: "refer to page 514"

Right: "refer to page 307"

(5) Page 87, S-Plus object (E).

The numbering in the S-Plus object (E) does not coincide exactly with that in the corresponding explanations. But the author believes that smart readers are not confused by this error.

(4) Page 75, Line 29 and 30.

Wrong: "Furthermore, since ${\bf A}_s$ does not depend upon $\lambda$, ${\bf U}_s$, which is used for the diagonalization of ${\bf A}_s$, does not depend upon $\lambda$ either."

Right: "Furthermore, since ${\bf S}$ does not depend upon $\lambda$, both of ${\bf U}_s$ and ${\bf A}_s$ do not depend upon $\lambda$ either."

(3) "PREFACE OF ENGLISH EDITION", 1st line

Wrong: "This book is the English edition of the first volume of Minnanotameno Nonparametorikku Kaiki, Second Edition (2003)"

Right: "This book is the English edition of Minnanotameno Nonparametorikku Kaiki, Second Edition (2003)"

(2) Page 516, Line 1 and 2 of (4).

Wrong: fit.pro <- ppr(xmat, yy, nterm=2, min.term = 1, max.term = 7, xpred = grid12)

Right: fit.pro <- ppr(xmat, yy, nterm=2, min.term = 1, max.term = 7)

This error does not affect the function of this object though.

(1) Title page

Wrong: "Tsukuba-shi ibaraki-ken, Japan"

Right: "Tsukuba-shi Ibaraki-ken, Japan"