I want to use gridsearch to get right parameters but got failed

[guanjen375]

I want to modify the parameter by gridsearchcv. Unforunately, I enforced an error when I want to do this:

Here is my code: (Same as /src/Train-Models/XGBoost_Model_ML.py)

import sqlite3
import pandas as pd
import xgboost as xgb
from sklearn.metrics import accuracy_score
from sklearn.model_selection import train_test_split
from tqdm import tqdm
import numpy as np
from sklearn import datasets, svm
from sklearn.model_selection import GridSearchCV

dataset = "dataset_2012-23"
con = sqlite3.connect("../../Data/dataset.sqlite")
data = pd.read_sql_query(f"select * from "{dataset}"", con, index_col="index")
con.close()

margin = data['Home-Team-Win']
data.drop(['Score', 'Home-Team-Win', 'TEAM_NAME', 'Date', 'TEAM_NAME.1', 'Date.1', 'OU-Cover', 'OU','TEAM_NAME','TEAM_NAME.1','TEAM_ID','TEAM_ID.1'],
axis=1, inplace=True)

data = data.values
data = data.astype(float)

x_train, x_test, y_train, y_test = train_test_split(data, margin, test_size=.1)

cv_param = {'max_depth': [2,3,4]}
param = {
'eta': 0.1,
'objective': 'multi:softprob',
'num_class': 2
}

clf = xgb.XGBClassifier(**param)
grid = GridSearchCV(estimator=clf,param_grid=cv_param)
print("Start Fit...")
grid.fit(x_train,y_train)

print("Generate best estimator...")
best_estimator = grid.best_estimator_
print(best_estimator)

The error message is as follows:
Start Fit...
ValueError: Classification metrics can't handle a mix of binary and multilabel-indicator targets

Can you tell me how to modify my code?

[kyleskom]

I think it has to do with objective': 'multi:softprob',

[guanjen375]

The parameters are:
'eta': 0.1,
'objective': 'multi:softprob',
'num_class': 2

I do not really know what you mean.