MachineLearning-Cover-Thumbnail

 

 

Errata if you purchased between  September 12  and November 6, 2019:

 

Page 21; Exercise 1.12: Replace 12% by 12.5% (This was kindly pointed out by Emilio Barone)

 

Page 41: 10-11 lines from end: replace “five” by “four” in two places (This was kindly pointed out by Emilio Barone)

 

Page 75: line 2: Replace “training set” with “test set” (This was kindly pointed out by Emilio Barone)

 

Page 95: six lines from end: replace 698.18 by 696.19 and 31.28 by 31.29 (This was kindly pointed out by Emilio Barone)

 

Page 96:  10 lines from end: replace “unconditional” by “conditional” (This was kindly pointed out by Emilio Barone)

 

Page 99: Figure 4.4: replace $353,593 by $353,573 (This was kindly pointed out by Emilio Barone)

 

Page 112: footnote 2: Replace  by  (This was kindly pointed out by Emilio Barone)

 

Page 113: Table 5.2 mistakenly calculates loans misclassified using the edges of the path rather than the middle.  When the middle of the path is used it is always the case that 0% are misclassified. This example will be changed for future printings. (I am grateful to Raymond Kan for pointing this out)

 

Page 117: second line above the figure: “C=0.1” should be “C=0.01”

 

Page 128: equation (6.2): insert a minus sign in front of the expression. Also the expression for the gradient in the final line of footnote 3 should be (E+E-)/0.02 (This was kindly pointed out by Emilio Barone)

 

Page 132: Caption for figure should read: Situation where there is a local minimum at B

 

Page 141: equation for  should have a minus sign in front of right hand side (This was kindly pointed out by Emilio Barone)

 

Page 172: Answer to 1.11: the standard deviation of the errors for the training set are correct, but those for the test set are wrong. They should be 35,588 for the polynomial of degree 3 model  and 37,427 for the polynomial of degree four model. (This was kindly pointed out by Silvio Bencini

 

Page 181: The table for Problem 5.11 mistakenly calculates loans misclassified using the edges of the path rather than the middle.  When the middle of the path is used, the loans misclassified for values of C equal to 0.01, 0.002, 0.001, 0.0005, 0.0003, 0.0002, and 0.0001 are 20%,20%, 20%, 20%, 30%, 40%, 40%, respectively.