从python中的xlsx文件中的一组行中获取随机项

Question

I have a xlsx file, for example:

A  B  C  D  E  F  G
1  5  2  7  0  1  8
3  4  0  7  8  5  9
4  2  9  7  0  6  2
1  6  3  2  8  8  0
4  3  5  2  5  7  9
5  2  3  2  6  9  1

being my values (that are actually on an excel file). I nedd to get random rows of it, but separeted for column D values.

You can note that column D has values that are 7 and values that are 2.

I need to get 1 random row of all the rows that have 7 on column D and 1 random row of all the rows that have 2 on column D.

And put the results on another xlsx file.

My expected output needs to be the content of line 0, 1 or 2 and the content of line 3, 4 or 5.

Can someone help me with that? Thanks!

Answer 1

With OpenPyXl, you can use Worksheet.iter_rows to iterate the worksheet rows.

You can use itertools.groupby to group the row according to the "D" column values. To do that, you can create a small function to pick-up this value in a row:

def get_d(row):
    return row[3].value

Then, you can use random.choice to choose a row randomly.

Putting all things togather, you can have:

def get_d(row):
    return row[3].value


for key, group in itertools.groupby(rows, key=get_d):
    row = random.choice(list(group))
    print(row)

Answer 2

I've created the code to that. The code below assumes that the excel name is test.xlsx and resides in the same folder as where you run your code. It samples NrandomLines from each unique value in column D and prints that out.

import pandas as pd
import numpy as np
import random
df = pd.read_excel('test.xlsx') # read the excel

vals = df.D.unique() # all unique values in column D, in your case its only 2 and 7

idx = []
N = []
for i in vals: # loop over unique values in column D
    locs = (df.D==i).values.nonzero()[0]
    idx = idx + [locs]  # save row index of every unique value in column D    
    N = N + [len(locs)] # save how many rows contain specific value in D



NrandomLines = 1 # how many random samples you want 

for i in np.arange(len(vals)): # loop over unique values of D
    for k in np.arange(NrandomLines): # loop how many random samples you want 
        randomRow = random.randint(0,N[i]-1) # create random sample

        print(df.iloc[idx[i][randomRow],:])  # print out random row