How to convert between Pandas DataFrame and NumPy ndarray

How can I convert Pandas DataFrame to NumPy array ?

When we use array data, we often use NumPy.

And pandas is also useful to handle table data.

We use both of them.

So sometimes we want to covert both format to each.

Today I introduce about "How to convert between Pandas DataFrame and NumPy ndarray".

Author

Mid-carieer engineer (AI, system). Good at Python and SQL.

How to convert Pandas DataFrame to NumPy ndarray

In order to convert Pandas DataFrame to NumPy ndarray, you can use values.

With Pandas DataFrame, try df.values.

Then you can get ndarray format data.

import numpy as np
import pandas as pd

# dataframe to numpy ndarray

data_list1 = [
[1,2,3],
[2,3,4],
[3,4,5],
[4,5,6]
]

col_list1 = ["c1","c2","c3"]
df1 = pd.DataFrame(data=data_list1, columns=col_list1)
print(df1)
print(type(df1))

#    c1  c2  c3
# 0   1   2   3
# 1   2   3   4
# 2   3   4   5
# 3   4   5   6
# <class 'pandas.core.frame.DataFrame'>

np_list = df1.values
print(np_list)
print(type(np_list))

# [[1 2 3]
#  [2 3 4]
#  [3 4 5]
#  [4 5 6]]
# <class 'numpy.ndarray'>

But according to Pandas document, values is not recommended.

They recommend to use to_numpy() instead.

np_list = df1.to_numpy()
print(np_list)
print(type(np_list))

# [[1 2 3]
#  [2 3 4]
#  [3 4 5]
#  [4 5 6]]
# <class 'numpy.ndarray'>

How to convert NumPy ndarray to Pandas DataFrame

Next how can we convert NumPy ndarray to Pandas DataFrame ?

In order to change ndarray to DataFrame, set ndarray as source data of DataFrame.

# numpy ndarray to dataframe

np_list = np.array(data_list1)
print(np_list)
print(type(np_list))

# [[1 2 3]
#  [2 3 4]
#  [3 4 5]
#  [4 5 6]]
# <class 'numpy.ndarray'>

col_list2 = ["c11","c22","c33"]
df2 = pd.DataFrame(data=np_list, columns=col_list2)
print(df2)
print(type(df2))

#    c11  c22  c33
# 0    1    2    3
# 1    2    3    4
# 2    3    4    5
# 3    4    5    6
# <class 'pandas.core.frame.DataFrame'>

It is same as converting list data to DataFrame.

If you want to add ndarray as new column, use df["column name"].

np_list2 = np.array([4,4,4,4])
df2["c44"] = np_list2
print(df2)
print(type(df2))

#    c11  c22  c33  c44
# 0    1    2    3    4
# 1    2    3    4    4
# 2    3    4    5    4
# 3    4    5    6    4
# <class 'pandas.core.frame.DataFrame'>

Conclusion

Today I explained about "How to convert between Pandas DataFrame and NumPy ndarray".

We can use below methods to convert.

Sometimes we use these. So write down them.