调整 XG Boost 参数时遇到问题

Question

我有一个分类问题，我选择使用 xg boost，它给了我很好的准确性，但是在使用随机搜索 cv iam 进行超参数调整时遇到了问题！

我制作的网格：

xg_grid = {"n_estimators":np.arange(1,10),
"max_depth": [None,6,8,10],
"learning_rate":[0.1,0.5,0.8,1],
"objective":"reg:logistic",
"max-depth": np.arange(6,10),
"alpha":np.arange(0,5),
"colsample_bytree":[0.1,1,0.5,0.3],
"booster":["gbtree","gblinear","dart"]}

适合它：

model =xgb.XGBClassifier(random_state=123)
rs_xg_boost = RandomizedSearchCV(model,
param_distributions=xg_grid,
cv=5,
n_iter = 10,
n_jobs = -1,
verbose=3)
rs_xg_boost.fit(x_train,y_train)

错误：

Fitting 5 folds for each of 10 candidates, totalling 50 fits

---------------------------------------------------------------------------
XGBoostError                              Traceback (most recent call last)
<ipython-input-98-f3a91cc91b88> in <module>
      6 n_jobs = -1,
      7 verbose=3)
----> 8 rs_xg_boost.fit(x_train,y_train)

~\Desktop\classifier_algorithm\env\lib\site-packages\sklearn\utils\validation.py in inner_f(*args, **kwargs)
     61             extra_args = len(args) - len(all_args)
     62             if extra_args <= 0:
---> 63                 return f(*args, **kwargs)
     64 
     65             # extra_args > 0

~\Desktop\classifier_algorithm\env\lib\site-packages\sklearn\model_selection\_search.py in fit(self, X, y, groups, **fit_params)
    878             refit_start_time = time.time()
    879             if y is not None:
--> 880                 self.best_estimator_.fit(X, y, **fit_params)
    881             else:
    882                 self.best_estimator_.fit(X, **fit_params)

~\Desktop\classifier_algorithm\env\lib\site-packages\xgboost\core.py in inner_f(*args, **kwargs)
    420         for k, arg in zip(sig.parameters, args):
    421             kwargs[k] = arg
--> 422         return f(**kwargs)
    423 
    424     return inner_f

~\Desktop\classifier_algorithm\env\lib\site-packages\xgboost\sklearn.py in fit(self, X, y, sample_weight, base_margin, eval_set, eval_metric, early_stopping_rounds, verbose, xgb_model, sample_weight_eval_set, feature_weights, callbacks)
    907             eval_group=None, label_transform=label_transform)
    908 
--> 909         self._Booster = train(xgb_options, train_dmatrix,
    910                               self.get_num_boosting_rounds(),
    911                               evals=evals,

~\Desktop\classifier_algorithm\env\lib\site-packages\xgboost\training.py in train(params, dtrain, num_boost_round, evals, obj, feval, maximize, early_stopping_rounds, evals_result, verbose_eval, xgb_model, callbacks)
    225     Booster : a trained booster model
    226     """
--> 227     bst = _train_internal(params, dtrain,
    228                           num_boost_round=num_boost_round,
    229                           evals=evals,

~\Desktop\classifier_algorithm\env\lib\site-packages\xgboost\training.py in _train_internal(params, dtrain, num_boost_round, evals, obj, feval, xgb_model, callbacks, evals_result, maximize, verbose_eval, early_stopping_rounds)
    100         # Skip the first update if it is a recovery step.
    101         if version % 2 == 0:
--> 102             bst.update(dtrain, i, obj)
    103             bst.save_rabit_checkpoint()
    104             version += 1

~\Desktop\classifier_algorithm\env\lib\site-packages\xgboost\core.py in update(self, dtrain, iteration, fobj)
   1278 
   1279         if fobj is None:
-> 1280             _check_call(_LIB.XGBoosterUpdateOneIter(self.handle,
   1281                                                     ctypes.c_int(iteration),
   1282                                                     dtrain.handle))

~\Desktop\classifier_algorithm\env\lib\site-packages\xgboost\core.py in _check_call(ret)
    187     """
    188     if ret != 0:
--> 189         raise XGBoostError(py_str(_LIB.XGBGetLastError()))
    190 
    191 

XGBoostError: [17:32:13] ..\src\objective\objective.cc:26: Unknown objective function: `i`
Objective candidate: survival:aft
Objective candidate: binary:hinge
Objective candidate: multi:softmax
Objective candidate: multi:softprob
Objective candidate: rank:pairwise
Objective candidate: rank:ndcg
Objective candidate: rank:map
Objective candidate: reg:squarederror
Objective candidate: reg:squaredlogerror
Objective candidate: reg:logistic
Objective candidate: reg:pseudohubererror
Objective candidate: binary:logistic
Objective candidate: binary:logitraw
Objective candidate: reg:linear
Objective candidate: count:poisson
Objective candidate: survival:cox
Objective candidate: reg:gamma
Objective candidate: reg:tweedie

我不明白有什么问题？ 我有我的网格并且我正确地安装了它，那为什么它不起作用？！

Answer 1

当为RandomizedSearchCV 提供一个网格作为dict 时，每个值都应该是要从中采样的参数列表，即使它只是一个。但是，您的网格包含以下键值对：

"objective": "reg:logistic"

这导致RandomizedSearchCV 对字符串"reg:logistic" 中的一个字符进行采样，而不是选择整个字符串。正确的做法是改为提供

"objective": ["reg:logistic"]