gedeck / mistat-code-solutions Goto Github PK

Describe the problem
The Granger causality test requires clarification why p-values are greater than the significance level. Statsmodels documentation states:

The Null hypothesis for grangercausalitytests is that the time series in the second column, x2, does NOT Granger cause the time series in the first column, x1. Grange causality means that past values of x2 have a statistically significant effect on the current value of x1, taking past values of x1 into account as regressors. We reject the null hypothesis that x2 does not Granger cause x1 if the pvalues are below a desired size of the test.

Suggested change
A clear and concise description of what you expected to happen.

Describe the problem

Chapter 2, Page 55, section 2.2.1.1. The bolded words after equation 2.13: "probability distribution function," shouldn't that be called "probability mass function"?

Change to code to create figures.

np.random.seed(1)
x = stats.norm(loc=10, scale=1).rvs(50)
fig, ax = plt.subplots(figsize=[5, 5])
pg.qqplot(x, ax=ax)
ax.get_lines()[0].set_color('grey')
ax.get_lines()[0].set_markerfacecolor('none')
ax.get_lines()[1].set_color('black')
ax.get_lines()[2].set_color('grey')
ax.get_lines()[3].set_color('grey')
plt.show()

Version 2.0 caused breaking changes to the code

Describe the problem
The pylibkriging package causes issues during installation. Find an alternative.

Describe the problem
File is missing for Industrial statistics

Consider adding --user to the pip install commands

Chapter 4, page 251, states that The square roots of these variances estimates are the "std err".... The Se value is shown in the regression summary output as Scale.
Page 250 states that Se^2 = 5.8869
Page 248 in the results summary states that Scale: 5.8832
Suspect there is rounding error, but none-the-less, the sqrt of 5.8832 = 2.426. Thus, Se =/= Scale

Problem:
The Scale output cannot be both variance and standard error.

https://www.statsmodels.org/dev/generated/statsmodels.regression.linear_model.OLSResults.scale.html
Note that the square root of scale is often called the standard error of the regression.

Solution:
The square roots of these variances estimates are the "std err".... The Se^2 value is shown in the regression summary output as Scale.

Describe the problem
Chapter 8 has deprecation warnings and failures.

Suggested change
Check and fix

Describe the problem
The pandas method iteritems is deprecated which causes the code in Example 7.3 to fail.

Suggested change

  df = pd.DataFrame([
    {satisfaction: counts for satisfaction, counts
      in response.value_counts().iteritems()},
    {satisfaction: counts for satisfaction, counts
      in response[q1_5].value_counts().iteritems()},
  ])

with

  df = pd.DataFrame([
    {satisfaction: counts for satisfaction, counts
      in response.value_counts().items()},
    {satisfaction: counts for satisfaction, counts
      in response[q1_5].value_counts().items()},
  ])

Describe the problem
Ask Springer to change this line on their website:

The mistat Python package can be accessed at https://gedeck.github.io/mistat-code-solutions/ModernStatistics/

Link should be clickable and should reference as source for code and solutions.

Suggested change
???

Describe the problem
Chapter 3 Page 162 Equation 3.30

The lower interval's denominator missing "/2" in the subscript of the Chi-square symbol. Should appear as: Chi-Square 1-a/2[n-1].

Describe the problem
The P-value explained at the top of page 152 but there is no reference to it in the index

Suggested change
Add "p. 152" to the P-value entry in the index on page 437.

Screenshots
If applicable, add screenshots to help explain your problem.

Software (if the problem is with the code):

• OS: [e.g. iOS]
• Python version [e.g. 3.10]

Additional context
Add any other context about the problem here.

• latex=True
• table formatting

Describe the problem
DtreeViz has a changed API

Suggested change
A clear and concise description of what you expected to happen.

Screenshots
If applicable, add screenshots to help explain your problem.

Software (if the problem is with the code):

• OS: [e.g. iOS]
• Python version [e.g. 3.10]

Additional context
Add any other context about the problem here.

https://iotespresso.com/how-to-install-graphviz-on-windows/

Describe the problem
The package used in the book seems to have low support. Check alternative GP models.

Suggested change
GPyTorch: https://docs.gpytorch.ai/en/latest/examples/01_Exact_GPs/Simple_GP_Regression.html
Scikit-learn: https://jekel.me/2016/Guassian-Process-Prediction-aka-Kriging/
PyKrige: https://geostat-framework.readthedocs.io/projects/pykrige/en/stable/contents.html

Describe the problem
Code has changed. New zip files are required. Also update the Install_packages notebook.

Describe the problem
Error below when using mistat.stepwise_regression(outcome=y, all_vars=X, data=df3)

Software Error
File ~\mistat\regression\stepwiseRegression.py:19, in find_best_model_partialF(outcome, variable_sets, data, old_model, opt_max)
17 with warnings.catch_warnings():
18 warnings.simplefilter("ignore")
---> 19 comparison = sms.anova.anova_lm(old_model, new_model)
20 if optF * partialF < optF * comparison.F[1]:
21 best_vars = variables

AttributeError: module 'statsmodels.stats' has no attribute 'anova'

Suggested change
import statsmodels.api as sm
---> 19 comparison = sm.stats.anova_lm(old_model, new_model)

Software (if the problem is with the code):
OS: Windows
Release: 10
Python implementation: CPython
Python version : 3.11.0
IPython version : 8.12.0
statsmodels: 0.14.0

Describe the problem
Example 9.17 Using the censored data from Exercise 9.16, we estimate the ....

should be:

Example 9.17 Using the censored data from Example 9.16, we estimate the ....

Describe the problem
Equation 3.30 is incorrect

Suggested change
The term $X_{1-\alpha}^2[n-1]$ needs to be changed to $X_{1-\alpha/2}^2[n-1]$

Make it clearer which book each directory refers to.

Describe the problem
Improve description of interaction plot figure.

Add information at the top of the files on what to do when Errors or Warnings occur.

• Warnings can usually be ignored
• Errors - google for the error and recommend fixes
• Check the repository for updated code

• Add date of last change to notebook exports
• Find way of having a last update date on the github.io pages

Make binder link similar to Modern Statistics

Add links like:

https://colab.research.google.com/github/gedeck/mistat-code-solutions/blob/main/IndustrialStatistics/notebooks/Chap002.ipynb

Example:
https://github.com/UVADS/DS1001/blob/master/ddsbook/analytics-lab-III.ipynb

Will require adding pip install statements at the start

Describe the problem
Error on page 135 in definition of W matrix constraints

Suggested change

For model_3 object in the code block, smf.ols() function formula argument has the wrong formula object referenced. Should read formula instead of poly_formula or vice versa.

Looking at the notebook at the following link: https://github.com/gedeck/mistat-code-solutions/blob/main/ModernStatistics/notebooks/Chap006.ipynb, it appears poly_formula is the selected object name.

This is just a object labeling inconsistency issue in the textbook.

Both books

Describe the problem
A clear and concise description of what the problem is.

Suggested change
A clear and concise description of what you expected to happen.

Screenshots
If applicable, add screenshots to help explain your problem.

Software (if the problem is with the code):

• OS: [e.g. iOS]
• Python version [e.g. 3.10]

Additional context
Add any other context about the problem here.

Some code breaks. Review all notebooks.

Proportional Sample Allocation Ch. 5 pg 317 Equation Clarification

The equation on page 317 in the top section under V$_{\text{simple}}$ ...

where
$\tilde{\sigma}_N^2 = \frac{N}{N-1}\sigma_N^2$

Should it read?:
$\tilde{\sigma}_{N_i}^2 = \frac{N_i}{N_i-1}\sigma_{N_i}^2$

I get the sense that it should because of how the equation for $\bar{\sigma}_N^2$ on page 318 is solved.

Add a page to guide readers on what they can do next (see predictive analytics FAQ for a start)

• Github copilot
• Chat-GPT2

Describe the problem
Deprecation warnings are not identified during tests. See if we can add code to identify warnings to address early on.