Computational Statistics

Errata and Clarifications

• Page 33, line 27 "wisely" should be
  "widely"
  It is possible in some cases that it is "wisely" used!
  (Thanks to Peter Tessin!)
• Page 51, line 8b, missing an alpha in the PDF of the Kumaraswamy distribution.
  The density is abx^(a-1)(1-x^a)^(b-1).
• Page 71, line 9b, instead of "Y" should be "h(y)".
• Page 78, Exercise 1.18(c) should be
  qgamma(0.5, 2, 1/3)
  yields 5.035
  (Thanks to Christophe Bartley!)
• Page 87, line 4: should be "2^{k-1}+i" (instead of "2^{k-1}-i"; i is negative).
  (Thanks to Andreas Eckner!)
• Page 100. In line 1 following Fig. 2.10, "4.992639e-09" should be "6.13826e-09", and
  in the following line, it should be "The solution is not even correct to 1 digit;".
  (Thanks to Todd Kuffner!)
• Page 109, lines 19 and 20: should be "the result in R may not exist in F" (instead of "the result in F may not exist in R").
  (Thanks again to Andreas Eckner!)
• Page 132. In lines 3 and 4, "x_{n-1}" should be "x_{n+1}", and
  in equation (3.17), "x_{k-1}" should be "x_{k+1}".
  (Thanks to Xiaochun Li!)
• Page 246. In step 5 of Algorithm 6.1, "s_1" should be "f_1", and
  equation (6.6) should be
  x^{(k+1)}=x^{(k)} -\frac{(f(x^{(k)}))^2}{f(x^{(k)}+f(x^{(k)}))-f(x^{(k)})}.
  (Thanks to Xiaochun Li!)
• Page 247. The formula in line 4b has a superflous ")"; that is, the formula should be the ceiling of the log divided by epsilon.
  (Thanks to Alexandre Leblanc!)
• Page 273. In step 4.a. of Algorithm 6.8, "x_a" should be "x_e"; that is, we should replace x_{m+1} by x_e.
  (Thanks to Alexandre Leblanc!)
• Page 407. In equation (10.8) and the following equation, "g(x)" should be "g(x_i)"
  (Thanks to Alex Meng!)
• Page 526, line 16 should say "top left-hand side".
  Page 526, line 19 should say "bottom left-hand side".
  (Thanks to Brad Patterson!)
• Page 536. The discussion of the cluster trees shown in Figure 16.11 is incorrect. (It was based on a different pair of trees that resulted from random data in which the seed was not set, so I cannot even reproduce it! The trees shown in Figure 16.11 are produced by the code shown on page 534 with using set.seed(3).)
  The tree shown on the left side in Figure 16.11 seems to suggest 3 clusters, {1,3}, {5}, and {2,4,6,7}. If we were to form 2 clusters in the tree shown on the right side in Figure 16.11, they would be {1,3,4} and {2,5,6,7}. (We also easily form 4 clusters in that tree: {3}, {1,4}, {5}, and {2,6,7}. As stated in the text, there is a certain amount of subjectivity in deciding on a number of clusters in a tree.) Using these two clusterings, the table in the middle of page 536 would have 2 rows and 3 columns, and the entries would be
  2 0 1
  0 1 3
  The row totals are 3 and 4, and the column totals are 2, 1, and 4.
• Page 537, line 6. Rand's statistic, based on the corrected table on page 536, is 8/21.
• Page 580, Exercise 16.1. Part b) should compare single linkage with average linkage.
  Parts c) and d) should compare average linkage with complete linkage. In part c) assume 4 clusters in average linkage, and in part d), assume 2 clusters in average linkage (and 3 clusters in complete linkage in both cases).
  (Thanks to Arun Vedachalam!)
• Page 678. In line 2 following the tables, "100,000" should be "90,000".
  (Thanks again to Todd Kuffner!)