Skip to content

Latest commit

 

History

History
199 lines (148 loc) · 6.38 KB

pandas.md

File metadata and controls

199 lines (148 loc) · 6.38 KB
Raw

Pandas Graphics

Pandas is a useful library for working with data. It relies on storing data in data frames. Instead of using a set of ordered pairs, a data frame in Pandas is similar to an Excel Spreadsheet or an SQL data frame and supports multiple rows and columns in a tabular fashion.

Visualization for Panda's data frames are an important tool for understanding the data set. Panda's visualization tools are based off of Matplotlib, so many of the plotting functions are the same.

Installation

To install Pandas, please use the command:

$ pip install pandas

Import Statements

The user will need to import both Pandas and Matplotlib in order to create and visualize data frames. In addition, Python's numpy may be a useful library for mathematical procedures on the data.

import matplotlib.pyplot as plt
import pandas as pd
import numpy as np

Bar Chart

Creating a bar chart with data frames is similar to creating bar charts with Matplotlib, with a couple differences in how you manipulate the data. In the following program, we use the same data and modifications as the Matplotlib bar chart example that can be found on Github.

import matplotlib.pyplot as plt
import pandas as pd

data = {'Rock': 136, 'Rap': 112, 'Folk': 110, 'Indie': 90, 'Jazz': 25}
categories = data.keys()
count = data.values()

df = pd.DataFrame({'Count':count, 'Categories':categories})

# Creating the bar chart
df.plot.bar(
        x='Categories',
        y='Count',
        align='center',
        color='orange',
        width=0.9,
        edgecolor="black",
        linewidth=2)

# Editing the bar chart's title, x, and y axes
plt.xlabel("Genre of Music")
plt.ylabel("Number of songs in the genre")
plt.title("Distribution of Genres in My Liked Songs")
plt.show()

This program can be downloaded from GitHub. The output of this program is showcased in @fig:pandas-barchart.

Pandas Bar Chart of Spotify Data{#fig:pandas-barchart width=50%}

Note the differences in creating the chart. Since data frames support multiple dimensions of data, the x and y we want to graph must be specified in df.plot.bar(). However, editing the title and axes are the same as in Matplotlib.

Line Chart

A line chart is typically used for time series data and non-categorical data. Pandas supports line chart visualization with plot.line(). We use the same data and modifications as the Matplotlib line chart example that can be found on GitHub. Note that since this data relies on random number generation the graphs will look slightly different each time.

import matplotlib.pyplot as plt
import pandas as pd
import random

x = []
y = []
for i in range(0, 100):
    x.append(i)
    value = random.random() * 100
    y.append(value)


df = pd.DataFrame({'x':x, 'y':y})

# creating the plot and labeling axes and title
df.plot.line(x='x', y='y')
plt.ylabel("y")
plt.title("Plot Test")

plt.show()

This program can be downloaded from GitHub. The output of this program is showcased in @fig:pandas-linechart.

Pandas Line Plot{#fig:pandas-linechart width=50%}

Pie Chart

A pie chart is useful for showing a division of a whole. Data that can be represented by a pie chart can also be used to make a bar chart, since both typically use categorical counts. Pandas uses plot.pie() to make a pie chart, in a manner similar to plot.bar(). We use the same data used to create the line chart for this visualization.

import matplotlib.pyplot as plt
import pandas as pd

data = {'Rock': 136, 'Rap': 112, 'Folk': 110, 'Indie': 90, 'Jazz': 25}
categories = data.keys()
count = data.values()

df = pd.DataFrame({'Count':count},index=categories)
plot = df.plot.pie(y='Count',legend=None)
save()

This program can be downloaded from GitHub.

The output of this program is showcased in @fig:pandas-piechart.

Pandas Pie Chart from Spotify Data{#fig:pandas-piechart width=50%}

Note that instead of listing both the Categories and the Count as data, we use the categories as index. This gets the proper labeling for our pie chart. In addition, Pandas automatically adds a legend, but this is unnecessary so we can remove the legend by setting the parameter legend=None in plot.pie().

Exporting

Note that this is the same as the Matplotlib tutorial found here on GitHub.

To export your graph as an image file, you can use the Matplotlib function savefig("fname.x"). You can specify the file type by filling in .x with .pdf, .png, svg, etc.

The parameter dpi sets the DPI (Dots per Inch) of the image being saved. Specify this number in the form of a float. For example, set dpi=300.

Additionally, there is another way to save files that may be faster than calling a specific method for each file. The following code showcases this:

import matplotlib.pyplot as plt
import os
from matplotlib import pyplot


def save():
    name = os.path.basename(__file__).replace(".py", "")
    plt.savefig(f'/filepath/{name}.png')
    plt.savefig(f'filepath/{name}.pdf')
    plt.savefig(f'filepath/{name}.svg')
    plt.show()

This code can be accessed on GitHub

The very last command is plt.show(), as this command displays the graph that you made. To show, simply type:

plt.show()

Links