Data Wrangling with Computational Notebooks: Glossary

Notebooks help connect the code for cleaning and wrangling data to the documentation explaining what is being done and why.

A jupyter notebook is divided into cells that are either code, markdown, or raw

Cells can be “run” leading to either the execution of code or formatting of markdown depending on the cell type

Code cells can be rerun, but this should be avoided to prevent obscuring the notebooks workflow

Pandas provides numerous attributes and methods that are useful for wrangling and analyzing data

Pandas contains numerous methods to help load/write data to/from files of different types

Select columns by using ["column name"] or rows by using the loc attribute

Sort based on values in a column by using the sort_values method

Using .dtypes to get the types of each column in a DataFrame

To get general statistics on the DataFrame you can use the describe method

You can add a constant to a numeric column by using the column + constant

Cleaning a dataset is an iterative process that can require multiple passes

Keep in mind to restart the kernel when cleaning a dataset to make sure that your code encompasses all the cleaning needed.

Grouping data by year and months is a powerful way to identify monthly and yearly changes

You can easily add more measurements to a single plot by using a list

There is a lot we didn’t cover here, so take a look at the Matplotlib docs (Link to Matplotlib docs) and other libraries that can allow you to make dynamic plots e.g. Plotly (Link to Plotly docs)