Scikit learn 在sklearn中，线性回归支持最多25个自变量？

当我使用sklearn做一些关于线性回归的事情时，如下所示：

x = df3.iloc[:,3:28].as_matrix().astype(int)
y = df3.iloc[:,0].as_matrix().astype(int)

from sklearn import linear_model
clf = linear_model.LinearRegression()

clf.fit(x,y)
print clf.coef_

print clf.score(x,y)

x = df3.iloc[:,3:29].as_matrix().astype(int)
y = df3.iloc[:,0].as_matrix().astype(int)

from sklearn import linear_model
clf = linear_model.LinearRegression()

clf.fit(x,y)
print clf.coef_

print clf.score(x,y)

它运行成功

[-55.06808051  -0.41350797   0.87904675   8.45228978  -3.54825048
  -8.63347841  22.26301155  20.82488927  -8.96890439  -8.67371986
  -6.99648938   0.07511078 -11.35819826  -7.51583817  -9.34964411
  10.41088005  12.30815737   8.55961149  14.69859856   9.81675829
  13.2959571   10.92224962   3.62143586  12.07015579   5.35530774]
0.455478830291

但如果我改变它如下：

x = df3.iloc[:,3:28].as_matrix().astype(int)
y = df3.iloc[:,0].as_matrix().astype(int)

from sklearn import linear_model
clf = linear_model.LinearRegression()

clf.fit(x,y)
print clf.coef_

print clf.score(x,y)

x = df3.iloc[:,3:29].as_matrix().astype(int)
y = df3.iloc[:,0].as_matrix().astype(int)

from sklearn import linear_model
clf = linear_model.LinearRegression()

clf.fit(x,y)
print clf.coef_

print clf.score(x,y)

它的结果是不切实际的：

[ -5.50139971e+01   8.30093647e+12   8.30093647e+12   8.30093622e+12
8.30093635e+12   8.30093635e+12   8.30093635e+12   8.30093635e+12
-1.09960450e+14  -1.09960448e+14  -1.09960450e+14  -1.09960448e+14
-1.09960450e+14  -1.09960448e+14  -1.09960450e+14  -1.09960450e+14
-1.09960444e+14  -1.09960450e+14  -1.09960448e+14  -1.09960450e+14
 1.50019494e+01   1.24627093e+01   5.25730696e+00   1.39210971e+01
 7.13531424e+00   8.90822183e+00]
-5841202768.85

为什么?？自变量的数量有限制吗

注：x已更改