In this short Python Pandas tutorial, we will learn how to convert a Pandas dataframe to a NumPy array. Specifically, we will learn how easy it is to transform a dataframe to an array using the two methods values and to_numpy, respectively. Furthermore, we will also learn how to import data from an Excel file and change this data to an array.
Now, if we want to carry out some high-level mathematical functions using the NumPy package, we may need to change the dataframe to a 2-d NumPy array.
Now, if we want to convert a Pandas dataframe to a NumPy array we need to have Python, Pandas, and NumPy installed, of course. Check the post about how to install Python packages to learn more about the installation of packages. It is recommended, however, that we install Python packages in a virtual environment. Finally, if we install and download a Python distribution, we will get everything we need. Nice and easy!
How do you convert a DataFrame to an array in Python?
To convert a Pandas DataFrame to a NumPy
array() we can use the values method (
DataFrame.to_numpy()). For instance, if we want to convert our dataframe called df we can add this code:
np_array = df.to_numpy().
Convert a Pandas Dataframe to a Numpy Array Example 1:
In this section, we are going to three easy steps to convert a dataframe into a NumPy array. In the first step, we import Pandas and NumPy. Step 2 involves creating the dataframe from a dictionary. Of course, this step could instead involve importing the data from a file (e.g., CSV, Excel). In the final step, we will use the values method to get the dataframe as an array.
Step #1: Import the Python Libraries
In the first example of how to convert a dataframe to an array, we will create a dataframe from a Python dictionary. The first step, however, is to import the Python libraries we need:
import pandas as pd import numpy as np
Step #2: Get your Data into a Pandas Dataframe
In the second step, we will create the Python dictionary and convert it to a Pandas dataframe:
Check the post about how to convert a dictionary to a Pandas dataframe for more information on creating dataframes from dictionaries. In the next step we are ready to change the dataframe to an array.
Step #3 Convert the Dataframe to an Array:
Finally, in the third step, we are ready to use the values method. Here’s how to convert the Pandas dataframe to a NumPy array:
# convert dataframe to numpy array df.values
That was easy, using the values method we converted the Pandas dataframe to a NumPy array in one line of code. In the next example, we are going to work with another method.
How to Change a Dataframe to a Numpy Array Example 2:
In the second example, we are going to convert a Pandas dataframe to a NumPy Array using the to_numpy() method. Now, the to_numpy() method is as simple as the values method. However, this method to convert the dataframe to an array can also take parameters.
Now, here’s a simple conversion example, generating the same NumPy array as in the previous the example;
# Pandas dataframe to numpy array: df.to_numpy()
If we want to convert just one column, we can use the dtype parameter. For instance, here we will convert one column of the dataframe (i.e., Share) to a NumPy array of NumPy Float data type;
# pandas to numpy only floating-point numbers: df['Share'].to_numpy(np.float64)
Note, if we wanted to convert only the columns containing integers we can use no.int64. For strings, we could input object. A final note, before going to the third example, is that is recommended to convert Pandas dataframe to an array using the to_numpy() method. In the next example, we are going to only select float and then convert the columns containing float values to a NumPy array.
Convert a Dataframe to a NumPy Array Example 3:
Now, if we only want the numeric values from the dataframe to be converted to NumPy array it is possible. Here, we need to use the select_dtypes method.
# Pandas dataframe to NumPy array selecting specific data types: df.select_dtypes(include=float).to_numpy()
Note, when selecting the columns with float values we used the parameter float. If we, on the other hand, want to select the columns with integers we could use int. Using this argument comes in handy when we want to e.g. calculate descriptive statistics or just want to extract certain data types from the NumPy array.
Read an Excel File to a Dataframe and Convert it to a NumPy Array Example 4:
Now, of course, many times we have the data stored in a file. For instance, we may want to read the data from an Excel file using Pandas and then transform it into a NumPy 2-d array. Here’s a quick an example using Pandas to read an Excel file:
# Reading the excel file df = pd.read_excel('http://open.nasa.gov/datasets/NASA_Labs_Facilities.xlsx', skiprows=1) # Exploring the first 5 rows and columns: df.iloc[0:5, 0:5]
Now, in the code, above we read an Excel (.xlsx) file from a URL. Here, the skiprows parameter was used to skip the first empty row. Moreover, we used Pandas iloc to slice columns and rows, from this df and print it. Here’s the result:
In the last example we will, again, use df.to_numpy() to convert the dataframe to a NumPy array:
# Converting the dataframe to an array: np_array = df.to_numpy()
Converting a Pandas dataframe to a NumPy array: Summary Statistics
In this last section, we are going to convert a dataframe to a NumPy array and use some of the methods of the array object. Again, we start by creating a dictionary. Second, we use the DataFrame class to create a dataframe from the dictionary. Finally, we convert the dataframe to a NumPy array only selecting float numbers.
Now that we have our NumPy array we can start using some methods for calculating summary statistics. First, we are going to summarize the two dimensions using the sum() method. Here’s an example code snippet:
# Summarizing the array np_array.sum(axis=0)
Second, we can calculate the mean values of the two dimensions using the mean():
# Calculating the mean of the array: np_array.mean(axis=0)
Note, that we used the parameter axis and set it to “0”. Now, if we didn’t use this parameter and set it to “0” we would have calculated it along each row, sort of speaking, of the array. This may be useful if we wanted to calculate the mean of scores across each observation in the dataset, for example. For example, if we have data from a questionnaire thought to measure different constructs, we may want to create a summary score for the complete scale (as well as for the constructs). In this case, we would remove the axis parameter.
DataFrame to Array YouTube Tutorial
Here’s also a YouTube Video explaining how to convert a Pandas dataframe to a NumPy array:
In this Pandas dataframe tutorial, we have learned how to convert Pandas dataframes to NumPy arrays. It was an easy task and we learned how to do this using values and to_numpy. As a final note, and as previously mentioned, you should use the later method for converting the dataframe.