Using Pandas

Install

When working with data, often within a DataFrame, the library of choice is Pandas. Install Pandas:

pip install pandas

Some OS already supply Pandas but before using the package check it is current:

pip install -U pandas

One can use Anaconda, but there are over 100 packages, so one can control the installation better by using Minicondas and install in a similar fashion to pip:

conda install pandas

There are three plugins that may require installing, dash, anyio and hypothesis. If working with large amounts of data it is recommended to install numexpr for accelerating some numerical operations, and bottleneck for certain nan evaluations.

Other dependancies may be necessary but as a start ensure matplotlib and numpy are present. If spreadsheet files are used for loading or saving xlrd, xlwt for xls files or openpyxl for reading and writing xlsx files. When working with statistics SciPy may be required, although pandas has a useful collection of built-in functions.

Seaborn can quickly visualise data. Some of its built in functions blend in very nicely with Pandas, many of its built-in tutorial examples are pandas' dataframes.

At a pinch the plotting functions of pandas itself can be used instead of matplotlib.

Using Pandas

One can work with scripts but it is eminantly suitable to use interactively. Some python IDEs (such as Idle and PyScripter) do not accept multiline inputs, check on your IDE of choice. If using Jupyter remember to start it within a directory that you have full control.

Open an OS command window, change to a user owned directory, then start with:

jupyter lab

This opens a user website in your browser. After the startup procedure is complete it starts in a console. Select the upper tab with +, then select the notebook.

Pandas works flexibly with the dataframes, if the command has no name (handle) then nothing changes, but the results of that command is shown.