Skip to main content

EDAV Fall 2021 Tues/Thurs Community Contributions

118 Plotting graph with R v.s. Python

Qi Meng

In this document, I’ll try to summerize some basic plots that we’ve learned in the course of STAT GR5702 Exploratory Data Analysis and Visualization and try to match the R codes with the corresponding Python codes. The examples provided in the document would be relatively simple. And I will introduce/use only some of the commonly used parameters in the examples. Hope the document can help people get some insights about which python packages/function to use if they want to create the plots.

The plots that I introduce in this document include Histogram, Boxplot, Density Curve, Ridgeline Plot, QQ Plot, Scatter Plot, Heatmap, Parallel Coordinate Plots and Mosaic Plot.

In this document, I referred to some codes used in the lecture and the problem sets. There are also some python codes that I referred online have been cited at the bottom of each section.

118.1 Histogram

118.1.1 R

We discussed two ways to create histogram in R, one uses the base R method and the other one uses the ggplot2.

118.1.1.1 Base R

  • The input data can be a column.
  • border: the color of histogram border.
  • col: the color to be filled in the histogram.
  • right: TRUE/FALSE value. TRUE stands for right-closed intervals, vise versa.
  • main: plot title.
  • xlab: label of x-axis.
x <- c(1, 5, 10, 20, 40, 50, 51, 53, 55, 56, 60, 65, 65, 68)
hist(x, border="blue", col = "lightblue", right = FALSE, main = 'Base R Histogram', xlab="data")

118.1.1.2 ggplot2

  • The input data should be a data frame. The column to be used in the histogram is specified in aes
  • geom_histogram is the one plotting histogram.
    • color: the color of histogram border.
    • fill: the color to be filled in the histogram.
    • binwidth: set the width of bin.
    • center: set the number to be a number that you want one of the column center on.
  • scale_x_continuous: set the scale of x-axis by setting its min value, max value and the interval value.
  • labs: set the labels and title for the plot
    • title: plot title.
    • x: label of x-axis.
    • y: label of y-axis.
x <- c(1, 5, 10, 20, 40, 50, 51, 53, 55, 56, 60, 65, 65, 68)
df <- data.frame(x)
ggplot(df, aes(x)) +
  geom_histogram(color = "blue", fill = "lightblue", binwidth = 10, center = 5) +
  scale_x_continuous(breaks = seq(0, 70, by = 5)) +
  labs(title = "ggplot Histogram", x = "data", y = "frequency")

118.1.2 Python

There are a few options/packages to plot histogram in python. I’ll use matplotlib.pyplot to plot the histogram in this example.

  • plt.subplot: a wrapper to create a fig and the axes for the plots to fit in.
  • plt.hist: the function to plot the histogram.
    • x: the data to be plotted.
    • bins: number of bins.
    • edgecolor: color of the border.
    • color: color filled in the histogram.
  • plt.set_title: plot title (subplot).
  • plt.set_xlable: label of x-axis (subplot).
  • plt.set_ylable: label of y-axis (subplot).
import matplotlib.pyplot as plt
import numpy as np
x = [1, 5, 10, 20, 40, 50, 51, 53, 55, 56, 60, 65, 65, 68]
df = np.array(x)
fig,ax = plt.subplots(1, 1, figsize=(5,5));
ax.hist(x=df, bins=7, edgecolor='blue', color='lightblue')
ax.set_title("Python Histogram")
ax.set_xlabel("data")
ax.set_ylabel("frequency")
plt.show()

118.2 Boxplot

118.2.1 R

We discussed two ways to create boxplot in R, one uses the base R method and the other one uses the ggplot2.

118.2.1.1 Base R

  • The input data can be a column.
  • main: plot title.
  • xlab: label of x-axis.
  • ylab: label of y-axis.
  • border: the color of boxplot border.
  • col: the color to be filled in the boxplot
x <- c(25, 50, 51, 53, 55, 56, 60, 65, 65, 68)
boxplot(x, main="Base R Boxplot", xlab="x label", ylab="y label", border="blue", col="lightblue")

118.2.1.2 ggplot2

  • The input data should be a data frame. The column to be used in the boxplot is specified in aes
  • geom_boxplot is the one plotting boxplot
    • color: the color of boxplot border.
    • fill: the color to be filled in the boxplot.
  • coord_flip: Flip cartesian coordinates.
x <- c(25, 50, 51, 53, 55, 56, 60, 65, 65, 68)
df <- data.frame(x)
ggplot(df, aes(x)) +
  geom_boxplot(color = 'blue', fill = 'lightblue') +
  coord_flip() +
  labs(title = "ggplot Boxplot", x = "x label", y = "y label")

118.2.2 Python

There are a few options/packages to plot boxplot in python. I’ll use matplotlib.pyplot to plot the boxplot in this example.

  • plt.boxplot: the function to plot the boxplot.
    • x: the data to be plotted.
    • vert: decide if we want a vertical boxplot.
x = [25, 50, 51, 53, 55, 56, 60, 65, 65, 68]
df = np.array(x)
fig,ax = plt.subplots(1, 1, figsize=(5,5));
ax.boxplot(x=df, vert=False)
ax.set_title("Python Boxplot")
ax.set_xlabel("x label")
ax.set_ylabel("y label")
plt.show()

118.3 Density Curve

118.3.1 R

I plot the density curve using ggplot2.

  • The input data should be a data frame. The column to be used in the density curve is specified in aes
  • geom_density is the one plotting density curve.
    • color: the color of density curve border.
    • fill: the color to be filled in the density curve
    • adjust: a multiplicate bandwidth adjustment, adjust=.5 means using half of the default bandwidth.
    • bw: the smoothing bandwidth to be used.
    • alpha: set the transparency for the plot.
x <- c(1, 5, 10, 20, 40, 50, 51, 53, 55, 56, 60, 65, 65, 68)
df <- data.frame(x)
ggplot(df, aes(x)) +
  geom_density(color='blue', fill="lightblue", adjust=.5, bw=5, alpha=.5) +
  labs(title = "ggplot Density Plot", x = "x label", y = "y label")

118.3.2 Python

There are a few options/packages to plot density curve in python. I’ll use seaborn to plot the density curve in this example.

  • plt.boxplot: the function to plot the density curve
    • x: the data to be plotted.
    • bw_adjust: a multiplicate bandwidth adjustment. Increasing the value would make the curve smoother.
    • fill: decide if fill the area under the curve with color.
    • ax: choose which ax to plot on. Can use the axes created by plt 
import seaborn as sns
x = [1, 5, 10, 20, 40, 50, 51, 53, 55, 56, 60, 65, 65, 68]
df = np.array(x)
fig,ax = plt.subplots(1, 1, figsize=(5,5));
sns.kdeplot(x, bw_adjust=.5, fill=True, ax=ax)
ax.set_title("Python Density Curve")
ax.set_xlabel("x label")
ax.set_ylabel("y label")
plt.show()

118.4 Ridgeline Plot

118.4.1 R

I plot the ridgeline plots using ggplot2.

  • The input data should be a data frame. The columns to be used in the ridgeline plots is specified in aes.
    • reorder: the function can be used to reorder the columns based on some values, such as median in the example below.
  • geom_density_ridges is the one plotting ridgeline plot.
    • color: the color of ridgeline plot border.
    • fill: the color to be filled in the ridgeline plot.
    • alpha: set the transparency for the plot.
ggplot(loans_full_schema, aes(x = loan_amount, y = reorder(loan_purpose, loan_amount, median))) +
  geom_density_ridges(color='blue', fill="lightblue", alpha=.5) +
  labs(title = "ggplot Density Plot", x = "x label", y = "y label")

118.4.2 Python

I’ll use joypy package to plot the ridgeline plot in this example. And I’ll use the preset iris dataset for displaying

  • joypy.joyplot: the function to plot the ridgeline plot
    • x: the data to be plotted.
    • by: when passing a column name to by, we will get a density plot for each value in the grouped column.
import joypy
import pandas as pd
iris = sns.load_dataset('iris')
fig, axes = joypy.joyplot(data=iris, by = 'species')
ax = plt.gca()
plt. show()

source: https://deepnote.com/@deepnote/Joyplot-Introduction-RmbhozJJRC6alCu8xcsbHQ

118.5 QQ Plot

118.5.1 R

I plot the QQ Plot plots using qqline. I’ll create 200 exponential data for displaying

  • qqnorm: produces a normal QQ plot of the values y passed in.
  • qqline is the one plotting QQ line
    • y: the data to be used to plot QQ line.
    • color: the color of QQ line.
y <- rexp(200, 5)
qqnorm(y=y)
qqline(y=y, col = "red")

118.5.2 Python

I’ll use scipy.stats package to plot the QQ plot in this example. I’ll create 200 exponential data for displaying

  • stats.probplot: the function to plot the QQ plot
    • x: the data to be plotted.
    • dist: distribution or distribution function name
    • plot: plots the quantiles if given
import scipy.stats as stats
y = np.random.exponential(scale = 5, size=200)
stats.probplot(x=y, dist="norm", plot=plt)
plt.show()

118.6 Scatter Plot

118.6.1 R

I plot the scatter plot using ggplot2. In the example below, we used the ames dataset from openintro library. The dependent variable would be area and the independent variable is price.

  • The input data should be a data frame. The columns(x/y) to be used in the scatter plot are specified in aes.
  • geom_point is the one plotting scatter plot.
    • size: the size of dot in the plot.
    • alpha: set the transparency for the plot.
data(ames)
ggplot(ames, aes(x = area, y = price)) + geom_point(size = 0.6, alpha = 0.2)

118.6.2 Python

There are a few options/packages to plot scatter plot in python. I’ll use matplotlib.pyplot to plot the scatter plot in this example. And I’ll generate 1000 random (x,y) data for displaying

  • plt.scatter: the function to plot the scatter plot
    • x: the data on x-axis to be plotted.
    • y: the data on y-axis to be plotted.
    • alpha: set the transparency for the plot.
x = np.random.rand(1000)
y = np.random.rand(1000)
fig,ax = plt.subplots(1, 1, figsize=(5,5));
ax.scatter(x, y, alpha=0.5)
ax.set_title("Scatter Plot")
ax.set_xlabel("x label")
ax.set_ylabel("y label")
plt.show()

118.7 Heatmap

118.7.1 R

I plot the heatmap using ggplot2. In the example below, we used the ames dataset from openintro library. The dependent variable would be area and the independent variable is price.

  • The input data should be a data frame. The columns(x/y) to be used in the heatmap are specified in aes.
  • geom_bin_2d is the one plotting heatmap.
    • bins: number of bins in both vertical and horizontal directions
data(ames)
ggplot(ames, aes(x = area, y = price)) + 
  geom_bin_2d(bins = 50)

118.7.2 Python

There are a few options/packages to plot heatmap in python. I’ll use seaborn to plot the heatmap in this example. And for the example below, I will use a preset dataset flights to display the heatmap in Python. * sns.heatmap: the function to plot the heatmap * data: a 2D dataset that can be coerced in to an ndarray. * ax: the ax that the heatmap would be plotted on.

fig,ax = plt.subplots(1, 1, figsize=(5,5));
df = sns.load_dataset("flights")
df = df.pivot("month", "year", "passengers")
sns.heatmap(data=df, ax=ax)
ax.set_title("Scatter Plot")
ax.set_xlabel("x label")
ax.set_ylabel("y label")
plt.show()

source: https://seaborn.pydata.org/generated/seaborn.heatmap.html

118.8 Parallel Coordinate Plots

118.8.1 R

In the lecture, we learned two type of R methods to plot the parallel coordinate plots. One uses ggparcoord which produces a static plot, while the other one uses parcoords to produce a interactive plot.

118.8.1.1 ggparcoord

I plot the parallel poordinate plots using ggparcoord. In the example below, we used the iris dataset. All the numerical columns are used below.

  • data: the dataset to plot.
  • column: the columns of the dataset to be used in the plot.
  • scale: method used to scale the variables. Some commonly used option is uniminmax and globalminmax.
  • title: the title of the graph.
  • alphaLines: the transparency of the line.
  • splineFactor: indicating whether spline interpolation should be used. The number will be multiplied by the number of columns.
ggparcoord(data=iris, column = 1:4, scale = 'globalminmax', title ="ggparcoord Parallel Coordinate Plots", alphaLines = 0.2, splineFactor = 10)

118.8.1.2 parcoords

I plot the parallel poordinate plots using parcoords. In the example below, we used the iris dataset. All the numerical columns are used below.

  • data: the dataset to plot.
  • `rowname``: the columns of the dataset to be used in the plot.
  • color: should include the list{colorScale=name of d3-scale, colorBy=the column that is used to determine the color, colorScheme=the color scheme to be used}.
  • alpha: the thickness of the line.
  • brushMode: the desired brush behavior.
  • withD3: whether or not include d3.js 
parcoords(data=iris, rowname=F, color=list(colorScale="scaleOrdinal", colorBy="Species", colorScheme="schemeCategory10"), alpha=0.5, brushMode='1d', withD3=TRUE)

118.8.2 Python

There are a few options/packages to plot parallel poordinate plots in python. I’ll use pandas to plot the parallel poordinate plots in this example. And for the example below, I will use a preset dataset iris to display the parallel poordinate plots in Python. * pd.plotting.parallel_coordinates: the function to plot the heatmap * frame: the dataset to plot * class_column: the column that is used to determine the color. * color: the colors to be used in the plot.

fig,ax = plt.subplots(1, 1, figsize=(5,5));
iris = sns.load_dataset("iris")
pd.plotting.parallel_coordinates(frame=iris, class_column="species", color=('#556270', '#4ECDC4', '#C7F464'))
ax.set_title("Parallel Poordinate Plots with Python")
ax.set_xlabel("x label")
ax.set_ylabel("y label")
plt.show()

source: https://pandas.pydata.org/docs/reference/api/pandas.plotting.parallel_coordinates.html

118.9 Mosaic Plot

118.9.1 R

In the lecture, we used vcd::mosaic to plot the mosaic plot. And I will show a simple mosaic plot with two variables only. The dataset I will used below is mpg.

  • formula: indicating the variables in data used to create a contingency table.
  • direction: indicating the direction of each variable
  • data: the dataset to plot.
  • highlighting_fill: the color to be filled in the tiles.
data(mpg)
vcd::mosaic(formula=drv~class, direction = c("v", "h"), data=mpg, highlighting_fill = c("lightblue", "lightpink", "lightyellow"))

118.9.2 Python

There are a few options/packages to plot mosaic plot in python. I’ll use statsmodels.graphics.mosaicplot to plot the mosaic plot in this example. And for the example below, I will use a preset dataset mpg to display the parallel poordinate plots in Python.

  • statsmodels.graphics.mosaicplot: the function to plot the mosaic plot
    • data: the dataset to plot.
    • index: indicating which variables/columns to be plotted.
from statsmodels.graphics.mosaicplot import mosaic
fig,ax = plt.subplots(1, 1, figsize=(5,5));
mpg = r.mpg
mosaic(data=mpg, index=['class', 'drv'])
ax.set_title("Parallel Poordinate Plots with Python")
ax.set_xlabel("x label")
ax.set_ylabel("y label")
plt.show()

source: https://data-science-master.github.io/lectures/09_python/09_matplotlib.html