Errata

In question 5 "You flipped a coin 19 times and got heads 15 times and tails 4 times." In appendix B question 5 "from scipy.stats import beta

heads = 8
tails = 2

p = 1.0 - beta.cdf(.5, heads, tails)

print(p) # 0.98046875". heads should be 15 and tails 4

Riemann Sums are spelled incorrectly as Reimann Sums

When applying the Bayes Theorem for the coffee/cancer problem, the P(Coffee) should be in the denominator for P(Cancer|Coffee). It is currently multiplying the numerator: P(Coffee|Cancer)*P(Coffee). The corrected version of the numerator should be P(Coffee|Cancer)*P(Cancer).

The order of operations paragraph omits the critical, but often forgotten, "from left to right." PEMDAS, while helpful, is only correct if one applies multiplication and division from left to right, and addition and subtraction from left to right. Omitting this renders the mnemonic unstable. The example on the following page says "The ordering of these two is swappable," which is technically incorrect. i.e. If you divide 25 by 5, you get 5, and if you multiply that by 2, you get 10, and if you subtract 4, you get the wrong answer. PEMDAS only works when multiplication and division is executed from left to right.

In the headline of example 1-8 and also in the description of the graph presented in Figure 1-3, the function x^2 + 1 is denoted as "exponential" function. To my knowledge, an exponential function is in the form y^x as e.g. exp(x), which can easily be generalized, while x^2 + 1 is a polynomial (function) - in this case a polynomial of second grade, also called a quadratic function.

62.6738 is more correctly expressed as 313369/5000 by sympy

The sentence "Conversely, you can turn an odds into a probability"... and then an example is shown. The example is turning a probability into odds. Hence, the sentence should be re-written as, "Conversely, you can turn a probability into an odds."

The labels for proabilities P(A) and P(B), i.e., Cancer and Coffee, are switched.

The coffee and cancer example accidentally switches the names for coffee and cancer variables in the written equation. P(Coffee) should be the denominator, not P(Cancer).

Bayes' Theorem formula is given as
P(A/B) = (P(B/A) x P(A)) / P(B)
which is correct.

But there is a mistake in a book while transforming the "coffee and cancer" problem according to Bayes' theorem.
The correct formula for the "probability of Cancer given the probability of coffee drinkers" is
P(Cancer|Coffee) = (P(Coffee|Cancer) * P(Cancer)) / P(Coffee)

But in the book, it is mentioned as
P(Cancer|Coffee) = (P(Coffee|Cancer) * P(Coffee)) / P(Cancer)

Whereas mathematical calculation is correct and it is according to the correct formula.

"We iterate each number of successes x" should read "We iterate each number of successes k."

Printed: The standard deviation for a sample and mean are specified by s and σ, respectively.

Should be: The standard deviation for a sample and population are specified by s and σ, respectively.

Inside the confidence interval function lower_ci should be subtracted from sample_mean but it is added in sample_mean which is wrong.

Corrected statement is
return sample_mean - lower_ci, sample_mean + upper_ci

The sentence "Since 16 is 4 days below the mean, we will also capture the area above 20, which is 4 days above the mean." The sentence should read "Since 16 is 2 days below the mean, we will also capture the area above 20, which is 2 days above the mean."

That paragraph says 16 is 4 days below the mean, but the mean is 18. 16 is 2 days below 18. It makes the same mistake in the other direction, saying 20 is 4 days above the mean. It's two.

The books states 0.5v = [3, 1.5]. When it should state 0.5v = [1.5,0.5]

values for i_hat and j_hat are swapped. i_hat should be [3,-1.5] and j_hat should be [2, -1].

The loglikelihood in the R^2 formula should be changed from -0.5596 to -14.341

Negative predicted value should be TN/(TN+FN) instead of TN/(TP+FN)

Author uses both sensitivity and recall in the figure, but fails to point out that they are the same thing.

Author provides code to demonstrate a confusion matrix, but does not include the output of the final print statement (which should print out a confusion matrix), which is surprising, since all previous examples do show the output of the print statements.

The sentence "Let's focus on finding the relationship on a weight from the output layer W_2 and the cost function C." Should perhaps read something like: ""Let's focus on finding the relationship on a weight (W_2) from the second output layer and the cost function C."

it should say W2, not W1