Python Functions Basics

Functions Overview

1. value_counts()
2. sort_values()
3. groupby()
4. agg()
5. pivot_table()
6. set_index()
7. loc[]
8. sort_index()

Detailed Breakdown

1. value_counts()

df['column_name'].value_counts(normalize=True, sort=True)

Key points:

• Used on a Series, not a DataFrame
• normalize=True gives proportions instead of counts
• sort=True is default, sorts by frequency descending

Common mistakes:

• Trying to use on entire DataFrame instead of a single column
• Forgetting that it returns a Series, not a DataFrame

2. sort_values()

df.sort_values(by='column_name', ascending=False)

Key points:

• Can sort by multiple columns using a list: by=['col1', 'col2']
• ascending=False for descending order
• Returns a new DataFrame unless inplace=True

Common mistakes:

• Forgetting to assign the result to a variable if not using inplace=True
• Using wrong column names

3. groupby()

df.groupby('column_name')['value_column'].operation()

Key points:

• Groups data by one or more columns
• Often followed by an aggregation function
• Can group by multiple columns: groupby(['col1', 'col2'])

Common mistakes:

• Forgetting to specify which column to perform operations on after grouping
• Not chaining an aggregation method after groupby()

4. agg()

df.groupby('column_name').agg({'col1': 'mean', 'col2': 'sum'})

Key points:

• Can apply multiple functions to different columns
• Works well with groupby()
• Can use custom functions or numpy functions

Common mistakes:

• Incorrect syntax when specifying multiple functions
• Forgetting to import numpy if using numpy functions

5. pivot_table()

pd.pivot_table(df, values='value_col', index='index_col', columns='column_col', aggfunc='mean', fill_value=0)

Key points:

• Creates a spreadsheet-style pivot table
• aggfunc defaults to mean if not specified
• Can use multiple columns for index or columns

Common mistakes:

• Confusing the roles of index, columns, and values parameters
• Forgetting to specify aggfunc when you want something other than mean

6. set_index()

df.set_index(['col1', 'col2'])

Key points:

• Creates a new index for the DataFrame
• Can create a MultiIndex by passing a list of column names
• Returns a new DataFrame unless inplace=True

Common mistakes:

• Forgetting to assign the result to a variable if not using inplace=True
• Not realizing that the columns used for the index are removed from the DataFrame body

7. loc[]

df.loc[row_indexer, column_indexer]

Key points:

• Uses labels/index values for selection
• Can select both rows and columns
• For slicing on MultiIndex, use tuples: df.loc[('country', 'city'):, :]

Common mistakes:

• Using integer positions instead of labels (that's what iloc[] is for)
• Incorrect syntax for MultiIndex slicing
• Forgetting the comma when selecting all rows or all columns (e.g., df.loc[:, 'column'])

8. sort_index()

df.sort_index(level=['col1', 'col2'], ascending=[True, False])

Key points:

• Sorts the DataFrame by the index
• For MultiIndex, use level parameter to specify which levels to sort by
• Can sort different levels in different directions using ascending parameter

Common mistakes:

• Confusing sort_index() with sort_values()
• Incorrect syntax for sorting multiple levels of a MultiIndex
• Forgetting that it returns a new DataFrame unless inplace=True

Remember, practice makes perfect! Keep working with these functions, and soon they'll become second nature. Happy coding!