Summary
Data processing and wrangling is the initial, and a very important, part of the data science pipeline. It is generally helpful if people preparing data have some domain knowledge about the data, since that will help them stop at the right processing point and use their intuition to build the pipeline better and more quickly. Data processing also requires coming up with innovative solutions and hacks.
In this chapter, you learned how to structure large datasets by arranging them in a tabular form. Then, we got this tabular data into pandas and distributed it between the right columns. Once we were sure that our data was arranged correctly, we combined it with other data sources. We also got rid of duplicates and needless columns, and finally, dealt with missing data. After performing these steps, our data was made ready for analysis and could be put into a data science pipeline directly.
In the next chapter, we will deepen our understanding of pandas and talk about reshaping and analyzing DataFrames for better visualizations and summarizing data. We will also see how to directly solve generic business-critical problems efficiently.
