我如何检查几乎所有行是否相同，如果是，则验证最后不同的行是否有一些常数 k

Question

using iris data set with dimension (150,4) I want to see if the rows are identical in columns 1 2 and 4 and if so then verify that the 3rd column has values differing by some constant k. This has to be done for every possible row combination.

#### load data###
import pandas as pd
df=pd.DataFrame(iris.data,columns=iris.feature_names)
df.head()

this gives me an error "The truth value of a Series is ambiguous. Use a.empty, a.bool(), a.item(), a.any() or a.all()." and also does not account for all the rows

I chose k=4 here

list=range(0,148)

for row in list:
    if df.iloc[row,:]==df.iloc[row+1,:]:
        df.iloc[row,2]-df.iloc[row+1,2]<=4
    else:
        print('nothing')

Answer 1

The error message basically tells you how to fix this:

if df.iloc[row,:]==df.iloc[row+1,:]:

it needs to be:

if (df.iloc[row,:]==df.iloc[row+1,:]).all():

Answer 2

First you need to filter the columns you want to compare. In this case columns 0, 1 and 3 by doing the following comparison df.iloc[row1,[0, 1, 3]] == df.iloc[row2, [0, 1, 3]] . This returns an array of True or False values. But you need all columns to be the same. To acomplish that you need the .all() method. It returns true only if all values in the array are True. In summary:

if (df.iloc[row1,identical_columns] == df.iloc[row2, identical_columns]).all():

And since you need to iterate over every posible row combination a double for loop will do nicely.

for row1 in range(m-1):
    for row2 in range(row1+1, m):
        # Check for every row combinaton if the columns are equal
        if (df.iloc[row1,identical_columns] == df.iloc[row2, identical_columns]).all():
            pass

Altogether:

import numpy as np
import pandas as pd

df = pd.DataFrame(data=np.random.randint(0, 100, (10, 4)))
m = df.shape[0]

identical_columns = [0, 1, 3]
k = 4

# Force rows values to pass
df.iloc[2, :] = [3, 4, 5, 1]
df.iloc[3, :] = [3, 4, 4, 1]

for row1 in range(m-1):
    for row2 in range(row1+1, m):
        # Check for every row combinaton if the columns are equal
        if (df.iloc[row1,identical_columns] == df.iloc[row2, identical_columns]).all():


            if df.iloc[row1,2] - df.iloc[row2,2] <= k:
                # TODO: Implement Your logic
                print ('We pass!')
        else:
            print(f"row {row1} and row {row2} don't pass")