Skip to content

Latest commit

 

History

History
248 lines (160 loc) · 6.77 KB

10_text_data.rst

File metadata and controls

248 lines (160 loc) · 6.77 KB

{{ header }}

.. ipython:: python

    import pandas as pd
Data used for this tutorial: 
  • .. ipython:: python

    titanic = pd.read_csv("data/titanic.csv")
    titanic.head()

How to manipulate textual data

  • Make all name characters lowercase.

    .. ipython:: python

    titanic["Name"].str.lower()

    To make each of the strings in the Name column lowercase, select the Name column (see the :ref:`tutorial on selection of data <10min_tut_03_subset>`), add the str accessor and apply the lower method. As such, each of the strings is converted element-wise.

Similar to datetime objects in the :ref:`time series tutorial <10min_tut_09_timeseries>` having a dt accessor, a number of specialized string methods are available when using the str accessor. These methods have in general matching names with the equivalent built-in string methods for single elements, but are applied element-wise (remember :ref:`element-wise calculations <10min_tut_05_columns>`?) on each of the values of the columns.

  • Create a new column Surname that contains the surname of the passengers by extracting the part before the comma.

    .. ipython:: python

    titanic["Name"].str.split(",")

    Using the :meth:`Series.str.split` method, each of the values is returned as a list of 2 elements. The first element is the part before the comma and the second element is the part after the comma.

    .. ipython:: python

    titanic["Surname"] = titanic["Name"].str.split(",").str.get(0)
    titanic["Surname"]

    As we are only interested in the first part representing the surname (element 0), we can again use the str accessor and apply :meth:`Series.str.get` to extract the relevant part. Indeed, these string functions can be concatenated to combine multiple functions at once!

To user guide

More information on extracting parts of strings is available in the user guide section on :ref:`splitting and replacing strings <text.split>`.

  • Extract the passenger data about the countesses on board of the Titanic.

    .. ipython:: python

    titanic["Name"].str.contains("Countess")


    .. ipython:: python

    titanic[titanic["Name"].str.contains("Countess")]

    (Interested in her story? See Wikipedia!)

    The string method :meth:`Series.str.contains` checks for each of the values in the column Name if the string contains the word Countess and returns for each of the values True (Countess is part of the name) or False (Countess is not part of the name). This output can be used to subselect the data using conditional (boolean) indexing introduced in the :ref:`subsetting of data tutorial <10min_tut_03_subset>`. As there was only one countess on the Titanic, we get one row as a result.

Note

More powerful extractions on strings are supported, as the :meth:`Series.str.contains` and :meth:`Series.str.extract` methods accept regular expressions, but are out of the scope of this tutorial.

To user guide

More information on extracting parts of strings is available in the user guide section on :ref:`string matching and extracting <text.extract>`.

  • Which passenger of the Titanic has the longest name?

    .. ipython:: python

    titanic["Name"].str.len()

    To get the longest name we first have to get the lengths of each of the names in the Name column. By using pandas string methods, the :meth:`Series.str.len` function is applied to each of the names individually (element-wise).

    .. ipython:: python

    titanic["Name"].str.len().idxmax()

    Next, we need to get the corresponding location, preferably the index label, in the table for which the name length is the largest. The :meth:`~Series.idxmax` method does exactly that. It is not a string method and is applied to integers, so no str is used.

    .. ipython:: python

    titanic.loc[titanic["Name"].str.len().idxmax(), "Name"]

    Based on the index name of the row (307) and the column (Name), we can do a selection using the loc operator, introduced in the :ref:`tutorial on subsetting <10min_tut_03_subset>`.

  • In the "Sex" column, replace values of "male" by "M" and values of "female" by "F".

    .. ipython:: python

    titanic["Sex_short"] = titanic["Sex"].replace({"male": "M", "female": "F"})
    titanic["Sex_short"]

    Whereas :meth:`~Series.replace` is not a string method, it provides a convenient way to use mappings or vocabularies to translate certain values. It requires a dictionary to define the mapping {from: to}.

Warning

There is also a :meth:`~Series.str.replace` method available to replace a specific set of characters. However, when having a mapping of multiple values, this would become:

titanic["Sex_short"] = titanic["Sex"].str.replace("female", "F")
titanic["Sex_short"] = titanic["Sex_short"].str.replace("male", "M")

This would become cumbersome and easily lead to mistakes. Just think (or try out yourself) what would happen if those two statements are applied in the opposite order…

REMEMBER

  • String methods are available using the str accessor.
  • String methods work element-wise and can be used for conditional indexing.
  • The replace method is a convenient method to convert values according to a given dictionary.
To user guide

A full overview is provided in the user guide pages on :ref:`working with text data <text>`.