Pandas writing list of numpy arrays from two different lists into CSV file-CodePudding

I have a list of two numpy arrays that look like these:

a_b =  [array([0.23078484, 0.23076418]), 
        array([0.3478484, 0.72076418]), 
        array([1.42590463, 1.42562456])] 

c_d = [array([0.23276474, 0.23276488]), 
       array([0.3498484, 0.72086418]), 
       array([1.43590464, 1.44562477])]

and I want to generate a csv file that looks like the following

Source       A         B          C 
a_b      0.23078484  0.3478484  1.42590463
a_b      0.23076418  0.72076418 1.42562456 
c_d      0.23276474  0.3498484  1.43590464
c_d      0.23276488  0.72086418 1.44562477

I have tried this so far

df = pd.DataFrame({'a_b': a_b , 'c_d': c_d}, columns = ['A', 'B','C'])
df.to_csv('test.csv', index=False)

But it gives this error

raise ValueError("All arrays must be of the same length")
ValueError: All arrays must be of the same length

CodePudding user response：

You could use np.hstack T to work convert the list of arrays to a numpy array; convert it to a DataFrame; then save it as a csv file:

out = (pd.DataFrame(np.hstack((a_b, c_d)).T, 
                    index=['a_b']*len(a_b[0]) ['c_d']*len(c_d[0]), 
                    columns=[*'ABC'])
       .reset_index()
       .rename(columns={'index':'Source'}))
out.to_csv('file.csv')

Another way that is similar to yours is to use the from_dict constructor with "orient" parameter. Then explode-ing the columns will get the desired outcome:

out = (pd.DataFrame.from_dict({'a_b': a_b , 'c_d': c_d}, 
                              orient='index', 
                              columns = ['A', 'B','C'])
       .explode(['A','B','C'])
       .reset_index()
       .rename(columns={'index':'Source'}))

Output:

  Source         A         B         C
0    a_b  0.230785  0.347848  1.425905
1    a_b  0.230764  0.720764  1.425625
2    c_d  0.232765  0.349848  1.435905
3    c_d  0.232765  0.720864  1.445625