Reputation: 21
I am trying to fit my xgboost model object (0.60 version) on OOT data, but keep getting error. I am using below line of code:
fname = "xgb"
if isinstance(xgb, XGBClassifier):
regressor = XGBClassifier()
r = pickle.load(open(fname, "rb" ))
print(r)
regressor._Booster = r._Booster
regressor.set_params(**r.get_xgb_params())
y_predict = regressor.predict(oot)
Error:
AttributeError: 'XGBClassifier' object has no attribute '_le'
I also tried scoring the OOT data using alternate way:
scored = scored_data.predict(oot)
Then i get below error (i have created similar environment replicating model dev)
class_probs = self.booster().predict(test_dmatrix,output_margin=output_margin,ntree_limit=ntree_limit)
TypeError: 'str' object is not callable
Upvotes: 2
Views: 3324
Reputation: 1958
I was getting the same issue with version 0.90. Upgrading to 1.6.1 fixed it for me.
Upvotes: 1