Category: Python
-
Pandas read_csv accepts a URL: Loading remote data files
Want to turn an online CSV file into a Python Pandas data frame? Just pass a URL to read_csv: Options such as usecols and index_col work! This works on all read_* methods, including read_excel, read_json, and read_parquet.
-
Saving a Pandas plot to a file with get_figure and savefig
After plotting a Python Pandas data frame in Jupyter, how do you save it to a file?
-
Pandas datetime index slicing: sort_index to avoid a KeyError
If your Python Pandas dataframe has a datetime index, use a slice to retrieve rows in a certain time period: But if the rows are unsorted? You get a KeyError. Solution:
-
Sorting a Python dict by value with operator.itemgetter
Sorting a Python dict by value? Instead of: Use itemgetter. It’s cleaner AND ~25% faster: Sort by value, then by key:
-
Python file modes: Use ‘x’ instead of ‘w’ to avoid overwriting
Write to a file in Python with open(filename, ‘w’). What if filename already exists? Its contents are gone. (I hope you have backups!) Instead, try: If filename already exists, ‘x’ raises an exception. For new files, ‘x’ and ‘w’ are the same.
-
Pandas nlargest: Finding top rows faster than sort_values
Retrieve rows of a Python Pandas data frame with the 6 largest values of column x: Even better: Why is it better? On a 3m row, 680 MB data frame, sort/tail took 622 ms. nlargest? Only 84 ms — 7x faster!
-
Pandas column arguments: Where one name works, a list does too
A Python Pandas rule of thumb: Wherever you can use a column name (string), you can use multiple column names (a list of strings):
-
Pandas sort_values: Sorting a data frame by multiple columns
To sort a Python Pandas data frame by column x: By x and y (ascending): By x and y, both descending: Ascending x, descending y:
-
Pandas select_dtypes: Selecting columns by their dtype
Need specific column types in Python Pandas? Use select_dtypes: