CODEWITHSUNDEEP
pythonpandas
user1569317
user1569317

Reputation: 2496

Conditionally iterate over columns in Pandas

I have a Pandas dataframe for which I would like to return the number of unique values in each column, except that some columns should be excluded.

This is how I am used to selecting unique values in a column, but I am not sure how to iterate it:

pd.unique(df.column_name.ravel())

My mind goes to something like this, but it obviously is not valid.

col_names = list(df.columns.values)
dont_include = ['foo', 'bar']
cols_to_include = [x for x in col_names if x not in dont_include]
for i in cols_to_include:
 col_unique_count = len(pd.unique(df.i.ravel())

What is the best solution?

Upvotes: 1

Views: 1170

Answers (1)

EdChum
EdChum

Reputation: 393983

Code can be simplified to this:

cols_to_include = df.columns[~df.columns.str.contains('foo')]
for col in cols_to_include:
  col_unique_count = df[col].nunique()

You can call nunique to get the count of unique values for a given Series

Or:

cols_to_include = df.columns[~df.columns.str.contains('foo')]
df[cols_to_include].apply(pd.Series.nunique)

here apply will call nunique on each column

EDIT

Use isin to test for membership and ~ to negate the boolean mask:

In [47]:
df = pd.DataFrame(columns = ['foo','baz','bar','pie'])
df

Out[47]:
Empty DataFrame
Columns: [foo, baz, bar, pie]
Index: []

In [48]:
dont_include = ['foo', 'bar']
cols = df.columns[~df.columns.isin(dont_include)]
cols

Out[48]:
Index(['baz', 'pie'], dtype='object')

You can then use my code as before to iterate over the sub-selection of your df

Upvotes: 3

Related Questions