How to randomly select rows from a dataset using pandas?

Question

I have a dataset with 36k rows. I want to randomly select 9k rows from it using pandas. How do you accomplish this task?

+3

Niranjan Agnihotri 28 Mar 17 at 6:56

2 answers

numpy

i = np.random.permutation(np.arange(len(df)))
idx = i[:9000]
pd.DataFrame(df.values[idx], df.index[idx])

+2

piRSquared 28 Mar 17 at 7:10

jezrael · Accepted Answer · 2017-03-28T06:58:35+0000

I think you can use sample

- 9k

or 25%

rows:

df.sample(n=9000)

Or:

df.sample(frac=0.25)

Another solution with making a random sample index

numpy.random.choice

and then choosing loc

- index

to be unique:

df = df.loc[np.random.choice(df.index, size=9000)]

Solution if not unique index:

df = df.iloc[np.random.choice(np.arange(len(df)), size=9000)]