Book Image

Python Data Cleaning Cookbook

By : Michael Walker
Book Image

Python Data Cleaning Cookbook

By: Michael Walker

Overview of this book

Getting clean data to reveal insights is essential, as directly jumping into data analysis without proper data cleaning may lead to incorrect results. This book shows you tools and techniques that you can apply to clean and handle data with Python. You'll begin by getting familiar with the shape of data by using practices that can be deployed routinely with most data sources. Then, the book teaches you how to manipulate data to get it into a useful form. You'll also learn how to filter and summarize data to gain insights and better understand what makes sense and what does not, along with discovering how to operate on data to address the issues you've identified. Moving on, you'll perform key tasks, such as handling missing values, validating errors, removing duplicate data, monitoring high volumes of data, and handling outliers and invalid dates. Next, you'll cover recipes on using supervised learning and Naive Bayes analysis to identify unexpected values and classification errors, and generate visualizations for exploratory data analysis (EDA) to visualize unexpected values. Finally, you'll build functions and classes that you can reuse without modification when you have new data. By the end of this Python book, you'll be equipped with all the key skills that you need to clean data and diagnose problems within it.
Table of Contents (12 chapters)

Looping through data with itertuples (an anti-pattern)

In this recipe, we will iterate over the rows of a DataFrame and generate our own totals for a variable. In subsequent recipes in this chapter we will use NumPy arrays, and then some pandas-specific techniques, for accomplishing the same tasks.

It may seem odd to begin this chapter with a technique that we are often cautioned against using. But I used to do the equivalent of looping every day 30 years ago in SAS, and on select occasions as recently as 7 years ago in R. That is why I still find myself thinking conceptually about iterating over rows of data, sometimes sorted by groups, even though I rarely implement my code in this manner. I think it is good to hold onto that conceptualization, even when using other pandas methods that work for us more efficiently.

I do not want to leave the impression that pandas-specific techniques are always markedly more efficient either. pandas users probably find themselves using apply...