Notes on Python

Installing Packages

There are two three ways: pip, conda and within Jupyter Hub. If you are using Anaconda (Navigator), use conda. pip will install for all environments; conda will install for conda environment only. Put either of these codes in Terminal.

# using pip
python -m pip install <package>

# using conda
conda install <package>

If it is a one off use case (which it usually never is), use the following command.

!pip install <package>

Terminal in Jupyter

Terminal commands can be passed within the Jupyter notebook, just like the pip install command described earlier. Any command starting with ! would be run like a terminal command within the Jupyter notebook.

# list all files in current directory
!ls

# delete all files in the current folder
!rm -r *

Debugging in the Current State

Many a time our code crashes and we would like to investigate the environment at that moment. Running %debug in Jupyter Notebook immediately after the code crashes would open a debugging environment where you can see variables' values, etc.

# some code that crashed
x = some_function(y)

# in the next Jupyter cell, run this command
%debug

Printing

There are two methods: print() and pprint(). They both serve the save cause but pprint() is better at displaying complex data structures such as list of lists or JSON files.

It is worthwhile to mention f-strings. They are usually more compact than writing full print statements. Furthermore, if you want to print value of a variable, they’re really short.

# enclose the variable in curly braces
print(f"Value of x is {x}")

# if you want to just print the value of, use {x=}
print(f"{x=}")

Adding Rows

It is more efficient to create a list of rows first and then convert it to a pandas data frame. As qmeeus said on SO,

Pandas dataframes do not work as a list, they are much more complex data structures and appending is not really considered the best approach.

data = []
for row in some_function_that_yields_data():
    data.append(row)

# either this
df = pd.DataFrame(data)

# or this
df = pd.concat([results, df], axis=0).reset_index(drop=True)

See Details of a Function

To see details of a function in a Jupyter notebook, use ?? operator. It will show you the body and the associated documentation (if available).

??my_function()