catboost_HPopt_CV

catboost_hp_cv.py

from catboost import Pool, cv, CatBoostClassifier
from bayes_opt import BayesianOptimization
from sklearn.model_selection import * 
from sklearn.metrics import *


def CB_opt(n_estimators, depth, learning_rate, max_bin,
             subsample, num_leaves, l2_leaf_reg, model_size_reg): 
  scores = []
  skf = StratifiedKFold(n_splits = 5, shuffle = True, random_state = 1944)
  for train_index, test_index in skf.split(X1, Y1):
    
    trainx, valx = X1.iloc[train_index], X1.iloc[test_index]
    trainy, valy = Y1.iloc[train_index], Y1.iloc[test_index]
 
    reg = CatBoostClassifier(verbose = 0,
                            n_estimators = int(n_estimators),
                            learning_rate = learning_rate,
                            subsample = subsample, 
                            l2_leaf_reg = l2_leaf_reg,
                            max_depth = int(depth),
                            num_leaves = int(num_leaves),
                            random_state = 88,
                            grow_policy = "Lossguide",
                            max_bin = int(max_bin),  
                            use_best_model = True, 
                            model_size_reg = model_size_reg,
                           
                            )
    
    reg.fit(trainx, trainy, eval_set = (valx, valy))
    scores.append(matthews_corrcoef(valy, reg.predict(valx)))
  return np.mean(scores)

pbounds = {"n_estimators": (150,400),
           "depth": (2,7),
           "learning_rate": (.01, 0.2),
           "subsample":(0.6, 1.),
           "num_leaves": (16,40),
           "max_bin":(150,300),
           "l2_leaf_reg":(0,10),
           "model_size_reg": (0,10)
}
optimizer = BayesianOptimization(
    f = CB_opt,
    pbounds = pbounds,
    verbose = 2,
    random_state = 888,
)

optimizer.maximize(init_points = 2, n_iter = 20)

print(optimizer.max)

Something's wrong with CB_opt function. Please check.
(What are X1 and Y1 in CB_opt function?)