autonomio / astetik Goto Github PK

If there are many values in label_col, the height of the graphic is not right.

This should be included in every plot where it makes sense.

The title says it all.

Several plots use _groubpy function in /utils/transforms.py to groupby various functions, including random, but that's not working.

Maybe in general scaling (to log and so forth) would be better to have two options; one where it's done on the plot level after creating the axes, and the other where it's done on the data before.

Full code and trace

https://colab.research.google.com/drive/1XBMF3tOQwRLuPrGJib-YkN20_9nmMVT4#scrollTo=Cgi6jl_6R6fj&line=8&uniqifier=1

most of the plots do not have a heading for the legend, which might cause confusion in many thinkable cases.

Instead of the fixed metrics right now (mean, std, total...) better to allow the user to choose which metrics they like.

There seems to be three options that the user should have:

• truncate (to a fixed length)
• rotate (does not work for legends)
• insert manually shorter ones (already added to line())

Instead of just the current 3 dv, there should be a way to add as many as one likes. The other option is to drop OLS entirely.

The groupby code can be made in to one function that serves both purposes.

maybe other plot too, but at least not in hist()

right now some of the plots have it as 'auto' even though it does not make sense to limit x for example when it's used for category labels (e.g. box())

I suppose that this has to do with the way the 'x' and 'y' are handled (and possibly confused with each other).

If line() gets multiple values in a list, scale is not working.

With the patient_data.csv file the x-labels are not showing correctly. Just integers are shown.

When doing import astetik the issue is with countries.csv. Will have to look for a better way to do this.

For some reason the seaborn backend return an error for this.

Hey,

first of all, thanks for creating and supporting talos/astetik! I came to astetik via Talos and both tools are highly useful in machine learning for someone not that advanced in coding like me.

While creating bargrid-plots i noticed, that the legend of the hue-variable often partially blocks some of the sub-plots (picture 1).

The docs to bargrid-function showed, that there was once a legend_position variable, which is currently commented out.

My question are:

• if it´s possible to (re-)enable the setting of the legend-position?
• or how/if it´s feasible to add an option with which it´s possible to change the legend-background to solid like you did in your machinelearningmastery article "Hyperparameter Optimization with Keras" (picture 2)

I think, that adding this functionality (back?) can greatly help the visualization, especially if you share it with other people.
If my questions are already implemented and just couldn´t find them i apologize. I´m quite new to python data visualization and problably didn´t notice the solutions.

Greetings

Right now it's standing out well. I think it will be good to standardize the legend across all plots to be as it's done in line() currently.

ast.hist() should allow two overlapping datasets simultaneously.

pip install astetik is failing because it is unable to install the deprecated sklearn dependency. In the pypi page of sklearn, there is a notice to start using scikit-learn instead of sklearn .
Request to change this in the requirements.txt and setup.py

Error message of pip installation in a fresh python environment

pip install astetik
Collecting astetik
  Using cached astetik-1.13-py3-none-any.whl (5.4 MB)
Collecting wrangle
  Using cached wrangle-0.7.2-py3-none-any.whl (52 kB)
Collecting pandas
  Using cached pandas-2.0.0-cp310-cp310-win_amd64.whl (11.2 MB)
Collecting geonamescache
  Using cached geonamescache-1.5.0-py3-none-any.whl (26.4 MB)
Collecting seaborn
  Using cached seaborn-0.12.2-py3-none-any.whl (293 kB)
Collecting statsmodels
  Using cached statsmodels-0.13.5-cp310-cp310-win_amd64.whl (9.1 MB)
Collecting sklearn
  Using cached sklearn-0.0.post4.tar.gz (3.6 kB)
  Preparing metadata (setup.py) ... error
  error: subprocess-exited-with-error

  × python setup.py egg_info did not run successfully.
  │ exit code: 1
  ╰─> [8 lines of output]
      Traceback (most recent call last):
        File "<string>", line 2, in <module>
        File "<pip-setuptools-caller>", line 34, in <module>
        File "C:\Users\mathavraj.j\AppData\Local\Temp\pip-install-bz3pwm5u\sklearn_08f62083d18c41bc8e66d870ee553b3f\setup.py", line 10, in <module>
          LONG_DESCRIPTION = f.read()
        File "C:\Users\mathavraj.j\AppData\Local\Programs\HCLTech\AION\2.7.0.3\python\lib\encodings\cp1252.py", line 23, in decode
          return codecs.charmap_decode(input,self.errors,decoding_table)[0]
      UnicodeDecodeError: 'charmap' codec can't decode byte 0x8f in position 7: character maps to <undefined>
      [end of output]

  note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

× Encountered error while generating package metadata.
╰─> See above for output.

note: This is an issue with the package mentioned above, not pip.```

Also markersize should be added as an option.

There seems to be some issue with the images on the main page of the docs:

I'm visualizing with Microsoft Edge 95.0.1020.53 64-bit

There should be a way to create the table dynamically in a way that allows any number of features to be included.

Currently the second color for the default palette is very light. Needs to be darker.

Some plots like multikde() are not using the standard scaling module. Such cases need to have a separate scaling function. Maybe it make sense to completely take out scaling from matplotlib and handle it all in numpy before passing the data. More options for scaling this way too.

Hello,

For some reason I get Indeted labels (different indention for different Scan runs). In the figure there is a lable indetion of the heatmap correlation figure for example.

How could this be solved?

Thank you

This seems to be just the case when there are very few values.