FE581 – Topics: R [Python and R for the ModernData Scientist]

Users often forget that Jupyter is shortfor “Julia, Python, and R” because it’s very Python-centric.

One last note: Python users refer to themselves as Pythonistias, which is a really coolname! There’s no real equivalent in R, and they also don’t get a really cool animal, butthat’s life when you’re a single-letter language. R users are typically called…wait forit…useRs! (Exclamation optional.) Indeed, the official annual conference is calleduseR! (exclamation obligatory), and the publisher Springer has an ongoing and veryexcellent series of books of the same name.

A python:R Bilingual Dictionary

View online with AzatAI Datalore Server

Package Management

Installing a single package

1
1
install.packages('tudyverse')

1
1
pip install pandas

Installing specific package versions

4
1
devtools::install_version(
2
"ggmap",
3
version = "3.5.2"
4
)

1
1
pip install pandas==1.1.0

Installing multiple packages

1
1
install.packages(c("sf", "ggmap"))

2
1
pip install pandas scikit-learn seaborn 
2
pip install -r requirements.txt

Loading packages

13
1
# Multiple calls to library()
2
library(MASS)
3
library(nlme)
4
library(psych)
5
library(sf)
6
# Install if not already available:
7
if (!require(readr)) {
8
install.packages("readr")
9
library(readr)
10
}
11
# Check, install if necessary, and
12
# load single or multiple packages:
13
pacman::p_load(MASS, nlme, psych, sf)

11
1
# Full package
2
import math
3
# Full package with alias
4
import pandas as pd
5
# Module
6
from sklearn import datasets
7
Module with alias
8
import statsmodels.api as sm
9
# Function
10
from statsmodels.formula.api import ols
11
# For ordinary least squares regression

Assign Operators

Fucking R has so much ugly assignments.

Types

The four most common user-defined atomic-vector types in R:

The four most common user-defined types in Python:

Arithmetic Operators

Common arithmetic operators:

Attributes

Class attributes:

​x
1
# List attributes
2
attributes(df)
3

4
# Access functions
5
dim(df)
6
names(df)
7
class(df)
8
comment(df)
9

10
# Add comment 
11
comment(df) <- "new info"
12

13
# Add custom attribute
14
attr(df, "custom") <- "alt info"
15
attributes(df)$custom

xxxxxxxxxx
14
1
# Definition of a class
2
class Food:
3
  name = 'toast'
4
  
5
# An instance of a class
6
breakfast = Food()
7

8
# An attribute of the class
9
# inherited by the instance
10
breakfast.name
11

12
# Setting an attibute
13
breakfast.name = 'simis'
14
# setattr(breakfast, 'name','simis')

Keywords

Reserved words and keywords:

Reserved words or keywords means you can not use them to name ur var.

xxxxxxxxxx
6
1
?reserved
2
if, else, repeat, while, function,
3
for, in, next, break, TRUE, FALSE,
4
NULL, Inf, NaN, NA, NA_integer_,
5
NA_real_, NA_complex_, NA_character_,
6
... (..1, ..2, etc.)

xxxxxxxxxx
12
1
# py Keywords
2
import keyword
3
print(keyword.kwlist)
4
## ['False', 'None', 'True', 'and',
5
'as', 'assert', 'async', 'await',
6
'break', 'class', 'continue', 'def',
7
'del', 'elif', 'else', 'except',
8
'finally', 'for', 'from', 'global',
9
'if', 'import', 'in', 'is', 'lambda',
10
'nonlocal', 'not', 'or', 'pass',
11
'raise', 'return', 'try', 'while',
12
'with', 'yield']

Functions and Methods

xxxxxxxxxx
5
1
# Basic definition
2
myFunc <- function(x, ...){
3
    x*10
4
}
5
myFunc(4)

40

xxxxxxxxxx
5
1
# Multiple unnamed arguments
2
myFunc <- function(...){
3
    sum(...)
4
}
5
myFunc(100,200,300)

600

xxxxxxxxxx
6
1
# Simple definition
2

3
def my_func(x):
4
    return x * 10
5

6
my_func(4)

xxxxxxxxxx
5
1
# Multiple named arguments, passed as a tuple
2
def my_func(*x):
3
    return x[2]
4

5
my_func(100,200,300)

300

xxxxxxxxxx
7
1
# Multiple unknown arguments, saved as a dict
2
def my_func(**num):
3
    print("x: ", num['x'])
4
    print("y: ", num['y'])
5

6

7
my_func(x=40, y=100)

x: 40 y: 100

Style and Naming Conventions

Style in R is generally more loosely defined than in Python. Nonetheless, see the Advanced R style guide by Hadley Wickham (CRC Press) or Google’s R Style guide forsuggestions.

For Python, see the PEP 8 style guide.

Analogous Data Storage Objects

Analogous Python objects for common R objects:

Analogous R objects for common Python objects:

xxxxxxxxxx
3
1
# Vectors
2
cities <- c("Istanbul", "Urumqi", "Almaty")
3
dist <- c(584,1054,653)

'Istanbul''Urumqi''Almaty' 584 1054 653

xxxxxxxxxx
3
1
# Lists
2
cities = ['Istanbul', 'Berlin', "Korla"]
3
dist = [584, 1054, 653]

['Istanbul', 'Berlin', 'Korla']

[584, 1054, 653]

One-dimensional, heterogeneous key-value pairs (Lists in R,dictionaries in Python):

xxxxxxxxxx
18
1
# A list of data frames
2
cities <- list(
3
    Munich = data.frame(
4
        dist=584,
5
        pop=143275,
6
        country="DE"
7
    ),
8
    Istanbul = data.frame(
9
        dist=5584,
10
        pop=1423275,
11
        country="TR"
12
    ),
13
    Almaty = data.frame(
14
        dist=84,
15
        pop=275,
16
        country="KZ"
17
    )
18
)

xxxxxxxxxx
2
1
# As a list object
2
cities[1]

xxxxxxxxxx
1
1
cities['Istanbul']

xxxxxxxxxx
2
1
# As a data.frame object (table)
2
cities[[1]]

A data.frame: 1 × 3

xxxxxxxxxx
1
1
cities$Almaty

A data.frame: 1 × 3

xxxxxxxxxx
4
1
# A list of heterogeneous data
2
lm_list <- lm(weight ~ group, data = PlantGrowth)
3
length(lm_list)
4
names(lm_list)

13 'coefficients''residuals''effects''rank''fitted.values''assign''qr''df.residual''contrasts''xlevels''call''terms''model'

xxxxxxxxxx
7
1
# lists
2

3
city = ['Munich', 'Paris', 'Amsterdam']
4
dist = [584, 1054, 653]
5
pop = [1484226, 2175601, 1558755]
6
area = [310.43, 105.4, 219.32]
7
country = ['DE', 'FR', 'NL'] 

xxxxxxxxxx
5
1
import numpy as np
2

3
# NumPy arrays
4
city_a = np.array(city)
5
city_a

xxxxxxxxxx
1
1
array(['Munich', 'Paris', 'Amsterdam'], dtype='%lt;U9')

xxxxxxxxxx
1
1
pop_a = np.array(pop)

xxxxxxxxxx
6
1
# Dictionaries 
2
yy = {'city': ['Munich', 'Paris', 'Amsterdam'],
3
      'dist': [584, 1054, 653],
4
      'pop': [1484226, 2175601, 1558755],
5
      'area': [310.43, 105.4, 219.32],
6
      'country': ['DE', 'FR', 'NL']}

xxxxxxxxxx
5
1
{'city': ['Munich', 'Paris', 'Amsterdam'],
2
 'dist': [584, 1054, 653],
3
 'pop': [1484226, 2175601, 1558755],
4
 'area': [310.43, 105.4, 219.32],
5
 'country': ['DE', 'FR', 'NL']}

Data Frames

Data Frames in Python:

xxxxxxxxxx
5
1
# class pd.DataFrame
2
import pandas as pd
3

4
# From a dictionary, yy
5
yy_df = pd.DataFrame(yy)

xxxxxxxxxx
6
1
# From lists
2
# names
3
list_names = ['city', 'dist', 'pop', 'area', 'country']
4

5
# columns are a list of lists
6
list_cols = [city, dist, pop, area, country]

['city', 'dist', 'pop', 'area', 'country']

[['Munich', 'Paris', 'Amsterdam'], [584, 1054, 653], [1484226, 2175601, 1558755], [310.43, 105.4, 219.32], ['DE', 'FR', 'NL']]

xxxxxxxxxx
6
1
# A zipped list of tuples
2
zip_list = list(
3
    zip(
4
        list_cols, list_names
5
    )
6
)

[(['Munich', 'Paris', 'Amsterdam'], 'city'), ([584, 1054, 653], 'dist'), ([1484226, 2175601, 1558755], 'pop'), ([310.43, 105.4, 219.32], 'area'), (['DE', 'FR', 'NL'], 'country')]

xxxxxxxxxx
1
1
zip_df = pd.DataFrame(zip_list)

xxxxxxxxxx
11
1
# Easier
2
# import pandas library
3
import pandas as pd
4

5
# init list of lists
6
list_rows = [['Munich', 584, 1484226, 310.43, 'DE'],
7
             ['Paris', 1054, 2175601, 105.40, 'FR'],
8
             ['Amsterdam', 653, 1558755, 219.32, 'NL']]
9

10
# Create the pandas data frame
11
df = pd.DataFrame(list_rows, columns=list_names)

Two-dimensional, heterogenous, tabular data frames in R:

xxxxxxxxxx
6
1
# class data.frame from vectors
2
cities_df <- data.frame(city = c("Munich", "Paris", "Amsterdam"),
3
dist = c(584, 1054, 653),
4
pop = c(1484226, 2175601, 1558755),
5
area = c(310.43, 105.4, 219.32),
6
country = c("DE", "FR", "NL"))

Multidimensional arrays:

R:

xxxxxxxxxx
6
1
# array
2

3
arr_r <- array(c(1:4,
4
seq(10, 40, 10),
5
seq(100, 400, 100)),
6
dim = c(2,2,3) )

xxxxxxxxxx
1
1
rowSums(arr_r,dims = 2)

xxxxxxxxxx
1
1
rowSums(arr_r,dims = 1)

444 666

xxxxxxxxxx
1
1
colSums(arr_r,dims = 1)

xxxxxxxxxx
1
1
colSums(arr_r,dims = 2)

Python:

xxxxxxxxxx
8
1
arr = np.array(
2
    [[[1, 2],
3
      [3, 4]],
4
     [[10, 20],
5
      [30, 40]],
6
     [[100, 200],
7
      [300, 400]]]
8
)

Logical Expressions

Relational operators

Python:

xxxxxxxxxx
1
1
import numpy as np
2
a = np.array([23,5,7,9,12])
3
a > 10 

xxxxxxxxxx
1
1
array([ True, False, False, False,  True])

Logical operators

xxxxxxxxxx
1
10
9
1
xx <- 1:10
2
xx == 6
3
xx !=6
4
xx >= 6
5
xx <- 1:6
6
# tails of a distribution
7
xx < 3 | xx >4 
8
# range in a distribution
9
xx > 3 & xx < 4

R for Pythonistas