scikit learn - python logistic regression - patsy design matrix and categorical data -

quite new python , machine learning.

i trying build logistic regression model. have worked in r gain lambda , used cross-validation find best model , moving python.

here have created design matrix , made sparse. ran logistic regression. seems working.

my question is, since have stated term item_number category how know has become dummy variable? , how know coefficient goes each category name?

from patsy import dmatrices sklearn.linear_model import logisticregression sklearn import preprocessing   def train_model (data, frm, rlambda):     y, x = dmatrices(frm , data, return_type="matrix")     y = np.ravel(y)      scaler = sklearn.preprocessing.maxabsscaler(copy=false)     x_trans = scaler.fit_transform(x)      model = logisticregression(penalty ='l2', c=1/rlambda)     model = model.fit(x_trans, y)  frm = 'purchase ~ price + c(item_number)' rlambda = 0.01 model, train_score = train_model(data1,frm,rlambda) 

first fix error code , answer question.

your code: train_model function won't return think returns. currently, doesn't return anything, , want return both model , training score. when fit model, need define mean training score - model won't return default. let's return model trained.

so should update train_model function follows:

def train_model (data, frm, rlambda):     y, x = dmatrices(frm , data, return_type="matrix")     y = np.ravel(y)      scaler = sklearn.preprocessing.maxabsscaler(copy=false)     x_trans = scaler.fit_transform(x)      model = logisticregression(penalty ='l2', c=1/rlambda)      # model.fit() operates in-place     model.fit(x_trans, y)      return model 

now when want determine variables correspond to, model.coef_ returns coefficients in decision function, of size (n_classes, n_features). order of coefficients correspond order features passed .fit() method. in case, x_trans design matrix of size (n_samples, n_features), each of coefficients in model.coef_ correspond coefficients each of n_features in x in same order presented in x.