Right now it works only on Python 2 (at least due to print statements, I didn't check if there are any bigger differences). Do you plan to make it Python 3 compatible?

introduce @property methods for

• get_constants
• get_categoricals
• get_numerics
• get_uniques
• get_bools

df = pd.DataFrame({
    'A': [1,2,3,4,5],
    'B': [0,1,1,1,0],
    'C': ['A', 'B', 'A', 'B', 'C'],
    'D': ['A', 'A', 'A', 'A', 'A'],
    'E': ['A', 'B', 'C', 'D', 'E']
})

dfs = DataFrameSummary(df)

print(dfs.get_numerics)
# Index(['A'], dtype='object')

print(dfs.get_bools)
# Index(['B'], dtype='object')

print(dfs.get_categoricals)
# Index(['C'], dtype='object')

print(dfs.get_constants)
# Index(['D'], dtype='object')

print(dfs.get_uniques)
# Index(['E'], dtype='object')

To address #10

Added from pandas.api import types.

Changed from common.is_numeric_dtype and common.is_datetime64_dtype in lines 113 and 116 to types.--- to make things work with pandas 0.23.0.

Not sure if you use from pandas.core import common anywhere else in your code but quick test with it removed doesn't seem to break anything.

Didn't do extensive testing but does seem to get things working again.

Should we think about merging this with pandas-profiling? Effort would be fairly low, and I like the idea of creating a DataFrameSummary object that you can query and which returns information.

Hello,

Thank you for all the work creating this project. It appears PyPI is considering 0.0.41 as the latest/default version instead of 0.0.5, therefore, 'pip install pandas-summary' is installing the old version.

Ryan

Could you please add the license to MANIFEST.in so that it will be included in sdists and other packages? This came up during packaging of pandas-summary in conda-forge.

This pull request is to add mode to the numeric statistic summary for columns as it is useful to know the most frequent value for some numeric columns, such as Age, Price, etc.

The Series implementation of mode() returns a series, hence the index reference - https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.Series.mode.html#pandas.Series.mode

Added mode to the testcase as well.

Also updated a testcase where the summary was getting re-sorted when it was already sorted. All test cases now pass.

Hi, I am trying to get pandas-summary working, and had version 0.0.41 installed. I ran into the issue fixed by #11/#12, so I tried to upgrade. It seems like the version was changed to 0.0.5 with https://github.com/mouradmourafiq/pandas-summary/commit/42227d51d8d458ebce7090db971259565fb6ccdf

When I try to upgrade to 0.0.5, pip picks up the 0.0.41 version since its patch level is considered greater than 0.0.5. I am somewhat new to the Python world, so I could be missing an easy way to do this, but wondering if the new version should be updated to 0.0.42 or 0.1.0 to better comply with semver.

pip install pandas-summary==
Collecting pandas-summary==
  Could not find a version that satisfies the requirement pandas-summary== (from versions: 0.0.3, 0.0.4, 0.0.5, 0.0.41)
No matching distribution found for pandas-summary==

For now my workaround is to use pip install pandas-summary==0.0.5 to install the exact version that I want to use.

Hi, I am trying to get pandas-summary working, and had version 0.0.41 installed. I ran into the issue fixed by #11/#12, so I tried to upgrade. It seems like the version was changed to 0.0.5 with https://github.com/mouradmourafiq/pandas-summary/commit/42227d51d8d458ebce7090db971259565fb6ccdf

When I try to upgrade to 0.0.5, pip picks up the 0.0.41 version since its patch level is considered greater than 0.0.5. I am somewhat new to the Python world, so I could be missing an easy way to do this, but wondering if the new version should be updated to 0.0.42 or 0.1.0 to better comply with semver.

pip install pandas-summary==
Collecting pandas-summary==
  Could not find a version that satisfies the requirement pandas-summary== (from versions: 0.0.3, 0.0.4, 0.0.5, 0.0.41)
No matching distribution found for pandas-summary==

For now my workaround is to use pip install pandas-summary==0.0.5 to install the exact version that I want to use.

I get the error

ValueError: supplied range of [0.0, inf] is not finite

when doing

DataFrameSummary(df)["OR"]

on a column with inf values.

Thanks for the library btw :)

I'd love to see these calculations split across multiple threads to reduce the clock computation time, since I often deal with really large data sets.

Thanks for the awesome plugin !

1. possible to add colors to point out missing values. Light shade of red if missing is > 0.
2. possible to display only missing columns? Sometimes a dataframe has a lot of columns and user is mostly interested in missing information.

I am new to python, if you guide me where to look, I can create a pull request. Thank you.

If a column has non-hashable objects, creation of DataFrameSummary gives an error:

TypeError: unhashable type: 'list'

It would be great if work also for these entries (e.g. looking at their length); or at very least - discard these variables.