Custom Sorting in Pandas Dataframes

In Pandas, sometimes you may need to sort a dataframe based on a custom order. This can be achieved using a dictionary to define the desired sorting order.

Problem:
You have a Pandas dataframe with a column containing month names. You want to sort this column using a custom dictionary, such as:

custom_dict = {'March':0, 'April':1, 'Dec':3}

Solution:

Using Categorical Series:
Pandas 0.15 introduced Categorical Series, which provides an elegant way to handle this scenario:

1. Convert the month column into a categorical series, specifying the custom ordering:

   df['m'] = pd.Categorical(df['m'], ["March", "April", "Dec"])

2. Sort the dataframe based on the categorical column:

   df.sort_values("m")

Using an Intermediary Series:
Prior to Pandas 0.15, you could utilize an intermediary series to achieve custom sorting:

1. Apply the custom dictionary to the month column:

   s = df['m'].apply(lambda x: {'March':0, 'April':1, 'Dec':3}[x])

2. Sort the intermediary series:

   s.sort_values()

3. Set the index of the dataframe to the sorted intermediary series and sort:

   df.set_index(s.index).sort()

Using the Replace Method:
In newer versions of Pandas, Series offers a replace method that allows for a more concise solution:

df['m'].replace({'March':0, 'April':1, 'Dec':3})

This method replaces the month values with the corresponding sorting values specified in the dictionary. Sorting the dataframe based on the modified month column will achieve the desired custom order.