How to fix AttributeError: 'Series' object has no attribute 'to_numpy'

Question

My output:

def load_data(self):
    """
    Load data from list of paths
    :return: 3D-array X and 2D-array y
    """
    X = None
    y = None
    df = pd.read_excel('data/Data.xlsx', header=None)
    for i in range(len(df.columns)):
        sentences_ = df[i].to_numpy().tolist()
        label_vec = [0.0 for _ in range(0, self.n_class)]
        label_vec[i] = 1.0
        labels_ = [label_vec for _ in range(0, len(sentences_))]
        if X is None:
            X = sentences_
            y = labels_
        else:
            X += sentences_
            y += labels_
    X, max_length = self.tokenize_sentences(X)
    X = self.word_embed_sentences(X, max_length=self.max_length)
    return np.array(X), np.array(y)

This is my code with pandas library as pd. When I run in Google Colab I get the following error:

AttributeError: 'Series' object has no attribute 'to_numpy'

Answer 1

Check the version of your pandas library:

import pandas
print(pandas.__version__)

If your version is less than 0.24.1:

pip install --upgrade pandas

Answer 2

If you need your code to work with all versions of pandas, here's a simple way to convert a Series into a NumPy array:

import pandas as pd
import numpy as np

s = pd.Series([1.1, 2.3])
a = np.array(s)
print(a)  # [1.1 2.3]

---

On an advanced note, if your Series has missing values (as NaN values), these can be converted to a masked array:

s = pd.Series([1.1, np.nan])
a = np.ma.masked_invalid(s)
print(a)  # [1.1 --]