The errata list is a list of errors and their corrections that were found after the product was released.
The following errata were submitted by our customers and have not yet been approved or disproved by the author or editor. They solely represent the opinion of the customer.
| Version |
Location |
Description |
Submitted by |
Date submitted |
| PDF |
Page Statistical Significance and p-Values
Code example |
In this section, we used the function perm_fun() which was defined in "Resampling" Section. The original function calculates the mean difference between samples, but in the example presented here, we need to calculate the difference between proportions. To solve this issue, a proposed solution is to create a new function that only changes this specific line of code:
return x.loc[list(idx_B)].mean() - x.loc[list(idx_A)].mean()
with this:
return x.loc[list(idx_B)].sum()/ nB - x.loc[list(idx_A)].sum()/ nA
(The code is in Python)
|
Anonymous |
May 02, 2023 |
| Printed |
Page AB testing
N/a |
I had a question about A/B testing which I hope you can help with.
My situation is that I am building a propensity model to identify which of our company's customers are most likely to sign up to a service after being sent an email. The model produces a likelihood score between 0 and 1 for each member and ranks them from 1 to X, where X is the size of our customer base. In practice we would then select the top N from this list to email. So, if our customer base is 1 million members, we would then select say the 100,000 ranked most highly (i.e. that have the highest likelihood score) by the model.
I want to compare how good the model is at identifying likely signups compared to the current business rules. However, I do not know how to conduct an A/B test in this scenario. Or indeed whether an A/B test is the most appropriate test here.
I understand the usual principle of randomly splitting the population into two groups and applying different treatments to each, such as a webpage layout or a drug treatment. However, in my case, the thing we are testing the efficacy of is the selection method itself (the email we send to each group would be the same). If we were to randomly split the population into two groups, then there will likely be customers who the model has ranked very highly, but which are in the business-rules group. Which seems to me like it wouldn't be fairly testing the model, because we are not giving it the chance to prove itself - we are not letting it have all of its 'top picks'.
Do you have any advice on this?
|
Anonymous |
Mar 19, 2024 |
| PDF |
Page 9
2nd paragraph |
Hi I do like to know what´s the exactly theme used in RStudio at the book practical statistics for data science.
I search for the theme but I couldn't find the exact theme in Rstudio as appear in the book. Aswell as the topography. It may be excellent if include that in a footnote in the same page you give the GitHub link.
It's seems without importance but I trying to practice the code specially in R language and I feel that if I use the same theme it would be easier to find my own mistakes. Somehow more pedagogy to me.
TIA,
PD: congratulations for the amazing book. I'm a native Spanish speaker and I didn't read an statistics related book with this degree of well writing and the explanation are accurate for those like me who are starting in this Data science universe.
-Albar
|
Albar Ugalde |
Oct 16, 2023 |
| Printed |
Page 53
Sample Mean Versus Population mean |
The symbol used to represent the mean of population is missing.
|
Anonymous |
Feb 07, 2022 |
| Printed |
Page 53
3rd paragraph |
left out symbol for mean of population in …”whereas is used to represent the mean of a population.”
|
John Taylor |
Jul 30, 2022 |
| Printed, PDF |
Page 99
4rd paragraph that start with "Page B has session times that are greater than those of page A by 35.67 seconds" |
i think in "The question is whether this difference is within the range of what random chance might produce, i.e., is statistically significant", last statement should removed "i.e., is statistically significant"
|
Mohammed Kamal Alsyd |
Jul 21, 2023 |
| Printed |
Page 137
3rd paragrapgh |
Because the earlier draws presented in both the previous page and in the page in question (137) are suggesting that whatever value (ones) Box A takes, the Box B gets as many zeros as the remainder of the whole, which is 10,000.
The whole being 10,000 and the Box B gets the remainder of what Box A got might not have been a strict rule there, but if it was meant to be as such, then from a standpoint of consistency, I suppose in the 3rd paragraph of page 137, with the boosted up new value of 165 (1.65%) ones for Box A, the Box B should also be equal to the remainder of 10,000 which would be 9835 (not 9868).
|
Emir Bilim |
Dec 24, 2021 |
| Printed |
Page 278
Inset on Ridge regression and the Lasso |
The indices of X should be X_{p,i} (cf. p. 151), but as it currently stands we have X_i and X_p. Shouldn't `i` refer to the example index and `p` refer to the dimension index?
|
Amine Laghaout |
Feb 15, 2023 |