CODEWITHSUNDEEP
pythonpandasdataframe
user987443
user987443

Reputation: 35

Using t.test_ind on specific columns of pandas dataframe

I have a pandas dataframe that is the format:

Variable    A1  A2  B1  B2  C1  C2  D1  D2

X   2   3   5   6   13  12  3   3

Y   1   1   7   9   16  19  11  9

Z   3   4   6   6   2   3   53  48

Where A1-A2, B1-B2, etc are replicate measurements and X, Y, Z are the different variables being measure.

I would like to do a t-test between D1-D2 and B1-B2 for each row and then append a new column with the p-values from each comparison.

Desired result would be:

Variable    A1  A2  B1  B2  C1  C2  D1  D2  p-val

X   2   3   5   6   13  12  3   3   0.0345

Y   1   1   7   9   16  19  11  9   0.111

Z   3   4   6   6   2   3   53  48  0.0004

Thank you in advance.

Upvotes: 0

Views: 723

Answers (2)

Rohith
Rohith

Reputation: 1048

I guess the method is right. But I am getting slightly different p-values

from scipy import stats
def my_func(x):
    x["p-val"] = stats.ttest_ind([x.D1,x.D2], [x.B1,x.B2]).pvalue
    return x
df = df.apply(my_func, axis=1)

Output:

Variable    A1  A2  B1  B2  C1  C2  D1  D2  p-val
X           2   3   5   6   13  12  3   3   0.037749551350623724
Y           1   1   7   9   16  19  11  9   0.29289321881345254
Z           3   4   6   6   2   3   53  48  0.0031413032318505603

Upvotes: 0

sacuL
sacuL

Reputation: 51335

I've got different results (I can't guess how you're doing your T-Test), but you can use scipy.stats.ttest_ind to do a t-test on independent variables, and extract the p-values from the result (the first index of the output, see linked doc for details):

from scipy.stats import ttest_ind

df['p-val'] = ttest_ind(df[['B1', 'B2']], df[['D1', 'D2']], axis=1)[1]

>>> df
  Variable  A1  A2  B1  B2  C1  C2  D1  D2     p-val
0        X   2   3   5   6  13  12   3   3  0.037750
1        Y   1   1   7   9  16  19  11   9  0.292893
2        Z   3   4   6   6   2   3  53  48  0.003141

Upvotes: 1

Related Questions