Data Reshaping with Pandas

Introduction

In data analysis, it is common to encounter datasets that are not arranged in the way we want. Reshaping data refers to the process of changing how data is organized into rows and columns. In this lab, we will explore some of the key methods in Pandas for reshaping data including Pivot, Stack, Unstack, Melt, Cross tabulations, Tiling, Factorizing, and Exploding.

VM Tips

After the VM startup is done, click the top left corner to switch to the Notebook tab to access Jupyter Notebook for practice.

Sometimes, you may need to wait a few seconds for Jupyter Notebook to finish loading. The validation of operations cannot be automated because of limitations in Jupyter Notebook.

If you face issues during learning, feel free to ask Labby. Provide feedback after the session, and we will promptly resolve the problem for you.

Importing Necessary Libraries and Create DataFrame

First, let's import the necessary libraries and create a DataFrame for our examples.

import pandas as pd
import numpy as np

## Create DataFrame
np.random.seed([3, 1415])
n = 20
cols = np.array(["key", "row", "item", "col"])
df = cols + pd.DataFrame((np.random.randint(5, size=(n, 4)) // [2, 1, 2, 1]).astype(str))
df.columns = cols
df = df.join(pd.DataFrame(np.random.rand(n, 2).round(2)).add_prefix("val"))

Pivoting with Single Aggregations

Pivot is one of the key methods for reshaping data in Pandas. It provides a way to transform your data so that you can view it from different angles.

## Pivot df with the mean of val0
df.pivot_table(values="val0", index="row", columns="col", aggfunc="mean", fill_value=0)

Pivoting with Multiple Aggregations

We can also perform multiple aggregations in Pivot.

## Pivot df with the mean and sum of val0
df.pivot_table(values="val0", index="row", columns="col", aggfunc=["mean", "sum"])

Cross Tabulations

Cross tabulation is a method to quantitatively analyze the relationship between multiple variables.

## Cross tabulation between row and col
df.pivot_table(index="row", columns="col", fill_value=0, aggfunc="size")

Summary

Congratulations! You have completed the Data Reshaping With Pandas lab. You can practice more labs in LabEx to improve your skills.