In [1]:
import numpy as np
import pandas as pd
from pandas import Series,DataFrame

In [2]:
np.random.seed(12345)

In [5]:
df = DataFrame(np.random.randn(1000,4))

df.head()


Out[5]:
0 1 2 3
0 -1.166530 -0.075264 0.112345 0.166874
1 0.012628 0.815313 -0.732001 0.868791
2 0.149693 0.485218 0.161056 -1.068808
3 1.190359 -1.053204 0.776001 1.311260
4 1.159677 0.477395 -0.004493 0.574631

In [6]:
df.tail()


Out[6]:
0 1 2 3
995 -1.530608 0.058991 -0.337120 1.129394
996 -0.151214 -0.898121 -1.467595 -1.086964
997 -0.104017 1.442214 -1.713617 0.936543
998 1.932427 0.530258 -0.217213 0.988951
999 0.193746 -1.744483 0.761804 -1.544201

In [7]:
df.describe()


Out[7]:
0 1 2 3
count 1000.000000 1000.000000 1000.000000 1000.000000
mean -0.039859 -0.033350 -0.011749 -0.031858
std 1.004647 0.998154 0.987471 1.013590
min -3.530912 -3.024110 -3.170292 -3.105636
25% -0.738643 -0.698731 -0.692887 -0.758010
50% 0.011608 -0.016171 0.022570 -0.035527
75% 0.663800 0.605952 0.666451 0.693491
max 2.916153 3.061029 2.623689 3.144389

In [10]:
col1 = df[0]

In [12]:
col1.head()


Out[12]:
0   -1.166530
1    0.012628
2    0.149693
3    1.190359
4    1.159677
Name: 0, dtype: float64

In [14]:
col1[np.abs(col1)>3]


Out[14]:
421   -3.530912
Name: 0, dtype: float64

In [15]:
df[(np.abs(df)>3).any(1)]


Out[15]:
0 1 2 3
158 0.623798 -0.436479 0.901529 -3.044612
192 0.617561 -1.148738 -3.170292 -1.017073
348 0.813014 -1.202724 -0.286215 -3.105636
360 0.123291 -3.024110 -1.168413 -0.888664
421 -3.530912 -0.576175 -0.750648 0.025443
712 -0.928871 3.061029 -0.297909 0.990886
787 0.836054 -0.780620 0.622791 3.144389

In [18]:
df[np.abs(df)>3] = np.sign(df) * 3

In [19]:
df.describe()


Out[19]:
0 1 2 3
count 1000.000000 1000.000000 1000.000000 1000.000000
mean -0.039328 -0.033386 -0.011579 -0.031852
std 1.002939 0.997895 0.986940 1.012700
min -3.000000 -3.000000 -3.000000 -3.000000
25% -0.738643 -0.698731 -0.692887 -0.758010
50% 0.011608 -0.016171 0.022570 -0.035527
75% 0.663800 0.605952 0.666451 0.693491
max 2.916153 3.000000 2.623689 3.000000

In [ ]:


In [ ]: