Not sure when this happened but i love the new left-hand side navigation https://scikit-learn.org/dev/user_guide.html

(@adrinjalali did this maybe?)

However, when clicking the different entries, the result is inconsistent. For some, it shows a TOC that expands the existing toc with m

Need to do some better handling of low-observation models in plot_diagnostics. These are models that shouldn't really be estimated, and we can't really make the plots work, but we shouldn't raise exceptions.

• Any dataset with less than 10 observations will raise an error computing the error autocorrelations:

mod = sm.tsa.statespace.SARIMAX(np.random.normal(size=10), order=(10, 

From: https://boltons.readthedocs.io/en/latest/setutils.html#boltons.setutils.IndexedSet

As you can see, the IndexedSet is almost like a UniqueList, retaining only one copy of a given value, in the order it was first added.

This appears incorrect. Perhaps it is a remnant from an earlier version of the library?

In tests it could be ok to have copy/paste code but I believe that in common estimator checks we could remove the boilerplate of datasets generation for each test by creating a common fixture. Apart from that we could reduce the number of instances of the dataset(s) in order to speed up the e

Hey there @JonnyWong16 , wanted to provide some content for the Wiki, and can't for the life of me figure out how to do a PR for Wiki content.

Below is an Nginx reverse proxy example that will allow users to access Tautulli via tautulli.url.com, and redirects HTTP requests to HTTPS. This configuration passes SSL Labs' test with an "A" grade.

DH Params creation vi

Currently the main statistics table looks as follows:

The following historical commit information, by author, was found in the repository:

Author           Commits    Insertions      Deletions    % of changes
author 1              24           564             59            10.2
author 2              21           276             97            30.2

It would be useful to add an ext

The figures in index.rst have URLs that don't work. I looked for them in the source tree, but I don't find them there.

The documentation of some classes/methods is severely lacking. Here's a list of methods that needs more detailed documentation as has been pointed out by users:

• TabularCPD doesn't clearly specify what the arguments are expected to be. Ref: #1036
• TabularCPD methods need to describe in more detail what each of them does.
• Make sure that Return section is available in ea

I'm testing out sybil locally to check all the "executable" blocks in the lifelines *.rst. This worked pretty well, I did discover some legit errors.

See #1011

• fix all "tests" in *.rst
• fix all "tests" in *.py
• make this part of TravisCI check

See mne-tools/mne-python#6262 (comment) :

seeing this in a example I would suggest to add a method Covariance.plot_topomap(info) to display the topomap of diagonal. this would simplify this example a lot

This will hopefully just require some short function in mne/viz/topomap.py. Even using the EvokedArray approach seems okay since it's pretty straigh

https://github.com/CamDavidsonPilon/lifetimes/blob/299229ebc6b01431f46979c4eb75362159134ddf/lifetimes/plotting.py#L251

Hi Cameron,

I've had feedback from customers that the output for plot_probability_alive_matrix() and plot_frequency_recency_matrix() functions is difficult to interpret.

The functions return plots with 'Customer Recency' on the y axis that appears to be a measure of t

Summary:

Documentation of optimizing does not appear accurate, specifically the init parameter

Description:

The description of the init parameter of optimizing is identical to that of sampling, with reference to chains, which optimizing does not use (as far as I understand).

Behavior does not seem to correlate with choices of init either. For example, initializing

1. Add sample size estimation documentation.
2. Improve multiple correction documentation adding all formulas and assumptions.

feel free to add more

For example,
in ELFI tutorial...
We should specify observed data like below:

Y = elfi.Simulator(MA2, t1, t2, observed=y_obs)

for calculating a discrepancy like:

d = elfi.Distance('euclidean', S1, S2)

What I wonder is,
without explicitly passing 'observed' argument to the Simulator object,
can we make it possible to do ABC-rejection by modifying discrepancy function?
e.g. inste

All functions are silent whether it be loading, preprocessing, or saving the data. Currently, the only step that prints to stdout is downloading. These other steps can somtimes be very expensive, such as the preprocessing in small32_imagenet.py.

We should consider adding

https://en.wikipedia.org/wiki/Local_outlier_factor

I noticed when making requests for Player information some of the NFL kickers seem to have a different format for their URI. A lot of current kickers follow the standard 8 character 'LlllFfNN' format. But some have a 10 character string.

In this example Mason Crosby this works fine and returns 'Field Goals - 30':

thisplayer = Player('CrosMa20') 
theyear = '2018'
print('Fie