我花了数小时试图使statsmodels进行MANOVA而不成功。 这是代码:
from statsmodels.multivariate.manova import MANOVA
df = data
feats_list = ['col1', 'col2', 'col3' ... 'col4']
var_list = ['col5', 'col6']
endog, exog = np.asarray(df[feats_list]), np.asarray(df[var_list])
manov = MANOVA(endog, exog)
manov.mv_test()
提供:
---------------------------------------------------------------------------
IndexError Traceback (most recent call last)
<ipython-input-16-c3fc1d1f16f6> in <module>()
1 manov = MANOVA(endog, exog)
----> 2 manov.mv_test()
~\Anaconda3\lib\site-packages\statsmodels\multivariate\manova.py in mv_test(self, hypotheses)
68 name = 'x%d' % (i)
69 L = np.zeros([1, self.exog.shape[1]])
---> 70 L[i] = 1
71 hypotheses.append([name, L, None])
72
IndexError: index 1 is out of bounds for axis 0 with size
1
我也尝试自己提出假设,并且总是收到SingularMatrixError,所以我想我没有正确使用该类。
预先感谢您的帮助。
答案 0 :(得分:0)
根据约瑟夫在上面的评论中引用的issue 4903,以下内容将起作用
from statsmodels.multivariate.manova import MANOVA
feats_list = ['col1', 'col2', 'col3', 'col4']
var_list = ['col5', 'col6']
df = pd.DataFrame(
np.random.random_sample(size=(100,6)),
columns=feats_list + var_list
)
endog, exog = np.asarray(df[feats_list]), np.asarray(df[var_list])
mod = MANOVA.from_formula('col1 + col2 + col3 + col4 ~ col5 + col6', data=df)
r = mod.mv_test()
print(r)
Multivariate linear model
============================================================
------------------------------------------------------------
Intercept Value Num DF Den DF F Value Pr > F
------------------------------------------------------------
Wilks' lambda 0.3420 4.0000 94.0000 45.2047 0.0000
Pillai's trace 0.6580 4.0000 94.0000 45.2047 0.0000
Hotelling-Lawley trace 1.9236 4.0000 94.0000 45.2047 0.0000
Roy's greatest root 1.9236 4.0000 94.0000 45.2047 0.0000
------------------------------------------------------------
------------------------------------------------------------
col5 Value Num DF Den DF F Value Pr > F
------------------------------------------------------------
Wilks' lambda 0.9297 4.0000 94.0000 1.7775 0.1399
Pillai's trace 0.0703 4.0000 94.0000 1.7775 0.1399
Hotelling-Lawley trace 0.0756 4.0000 94.0000 1.7775 0.1399
Roy's greatest root 0.0756 4.0000 94.0000 1.7775 0.1399
------------------------------------------------------------
------------------------------------------------------------
col6 Value Num DF Den DF F Value Pr > F
------------------------------------------------------------
Wilks' lambda 0.9891 4.0000 94.0000 0.2590 0.9035
Pillai's trace 0.0109 4.0000 94.0000 0.2590 0.9035
Hotelling-Lawley trace 0.0110 4.0000 94.0000 0.2590 0.9035
Roy's greatest root 0.0110 4.0000 94.0000 0.2590 0.9035
============================================================