sklearn and statsmodels getting very different logistic regression results

Question

from sklearn.linear_model import LogisticRegression
from io import StringIO
import statsmodels.api as sm

TESTDATA = StringIO("""a;b;c
1;8;5
0;5;4
0;4