[Tut] How to Convert a CSV to NumPy Array in Python?

[xSicKxBot]

How to Convert a CSV to NumPy Array in Python?

5/5 – (1 vote)

Problem Formulation

Given a CSV file (e.g., stored in the file with name

Code:

'my_file.csv'

).

Code:

INPUT: file 'my_file.csv' 9,8,7 6,5,4 3,2,1

Challenge: How to convert it to a NumPy Array?

Code:

OUTPUT: 2D NumPy Array [[9. 8. 7.] [6. 5. 4.] [3. 2. 1.]]

Method 1: np.loadtxt()



np.loadtxt()

You can convert a CSV file to a NumPy array simply by calling

Code:

np.loadtxt()

with two arguments: the

Code:

filename

and the

Code:

delimiter

string. For example, the expression

Code:

np.loadtxt('my_file.csv', delimiter=',')

returns a NumPy array from the

Code:

'my_file.csv'

with delimiter symbols

Code:

','

.

Here’s an example:

Code:

import numpy as np array = np.loadtxt('my_file.csv', delimiter=',') print(array)

Output:

Code:

[[9. 8. 7.] [6. 5. 4.] [3. 2. 1.]]

Method 2: np.loadtxt() with Header



np.loadtxt() + header

You can convert a CSV file with first-line header to a NumPy array by calling

Code:

np.loadtxt()

with three arguments: the

Code:

filename

,

Code:

skiprows=1

to skip the first line (header), and the

Code:

delimiter

string. For example, the expression

Code:

np.loadtxt('my_file.csv', skiprows=1, delimiter=',')

returns a NumPy array from the

Code:

'my_file.csv'

with delimiter symbols

Code:

','

while skipping the first line.

[i]Figure: Skip the first header line in the CSV using the

Code:

skiprows

argument of the

Code:

np.loadtxt()

function.[/i]

Here’s an example:

Code:

import numpy as np array = np.loadtxt('my_file.csv', skiprows=1, delimiter=',') print(array)

Output:

Code:

[[9. 8. 7.] [6. 5. 4.] [3. 2. 1.]]

Method 3: CSV Reader



CSV Reader

To convert a CSV file

Code:

'my_file.csv'

into a list of lists in Python, use the

Code:

csv.reader(file_obj)

method to create a CSV file reader. Then convert the resulting object to a list using the

Code:

list()

constructor. As a final step, you can convert the nested list to a NumPy array by using the

Code:

np.array(list)

constructor.

Here’s an example:

Code:

import numpy as np import csv csv_filename = 'my_file.csv' with open(csv_filename) as f: reader = csv.reader(f) lst = list(reader) print(lst)

The output is the list of lists:

Code:

[['9', '8', '7'], ['6', '5', '4'], ['3', '2', '1']]

Now, if you need to convert it to a NumPy array, you can simply use the

Code:

np.array()

function on the newly-created list like so:

Code:

array = np.array(lst) print(array)

Output:

Code:

[['9' '8' '7'] ['6' '5' '4'] ['3' '2' '1']]

Related Tutorial: How to Convert CSV to List of Lists in Python

Method 4: np.genfromtxt()



np.genfromtxt()

You can convert a CSV file to a NumPy array simply by calling

Code:

np.genfromtxt()

with two arguments: the

Code:

filename

and the

Code:

delimiter

string. For example, the expression

Code:

np.genfromtxt('my_file.csv', delimiter=',')

returns a NumPy array from the

Code:

'my_file.csv'

with delimiter symbol

Code:

','

.

Here’s an example:

Code:

import numpy as np array = np.loadtxt('my_file.csv', delimiter=',') print(array)

Output:

Code:

[[9. 8. 7.] [6. 5. 4.] [3. 2. 1.]]

Method 5: Pandas read_csv() and df.to_numpy()



read_csv() and df.to_numpy()

A quick and efficient way to read a CSV to a NumPy array is to combine Pandas’

Code:

pd.read_csv()

function to read a given CSV file to a DataFrame with the

Code:

df.to_numpy()

function to convert the Pandas DataFrame to a NumPy array.

Here’s an example:

Code:

import pandas as pd df = pd.read_csv('my_file.csv', header=None) array = df.to_numpy() print(array)

Output:

Code:

[[9 8 7] [6 5 4] [3 2 1]]

Related Tutorial: 17 Ways to Read a CSV File to a Pandas DataFrame

Summary

We have seen five ways to convert a CSV file to a 2D NumPy array:

• Method 1:
  Code:

  np.loadtxt()
  
  
• Method 2:
  Code:

  np.loadtxt()
  with Header
  
  
• Method 3: CSV Reader
  
  
• Method 4:
  Code:

  np.genfromtxt()
  
  
• Method 5: Pandas
  Code:

  read_csv()
  and
  Code:

  df.to_numpy()

Our preferred way is

Code:

np.loadtxt()

for its simplicity and Pandas for its extensibility.