2

I wanna understand why in this code, I get the following results:


# Import necessary libraries
import pandas as pd
from sklearn.metrics import f1_score
from sklearn.model_selection import train_test_split
from sklearn.metrics import f1_score, confusion_matrix
import lightgbm as lgb
import optuna
import mlflow
import mlflow.lightgbm

Load Titanic dataset


titanic_data = pd.read_csv('titanic.csv')  # Assuming the dataset is stored in 'titanic.csv'


Select specific features


selected_features = ['Pclass', 'Sex', 'Age', 'SibSp', 'Parch', 'Fare', 'Embarked', 'Survived']
titanic_data = titanic_data[selected_features]


Convert categorical features to numerical using one-hot encoding


titanic_data = pd.get_dummies(titanic_data, columns=['Sex', 'Embarked'], drop_first=True)


Extract features and target variable


X = titanic_data.drop('Survived', axis=1)
y = titanic_data['Survived']


Split dataset


X_train, X_valid, y_train, y_valid = train_test_split(X, y, test_size=0.2, random_state=42)


Train an initial LightGBM model


initial_model = lgb.train({}, lgb.Dataset(X_train, label=y_train), 100)


Make predictions and calculate F1 score


y_pred_initial = initial_model.predict(X_valid)
f1_initial = f1_score(y_valid, (y_pred_initial > 0.5).astype(int))


Display F1 Score and Confusion Matrix


print(f"Initial F1 Score: {f1_initial}")
print("Confusion Matrix:")
print(confusion_matrix(y_valid, (y_pred_initial > 0.5).astype(int)))

When printing y_pref_initials I get an array of probabilities that has negative values and values greater than 1 : Example:


array([ 0.01546079,  0.17557856,  0.22971758,  1.23292351,  0.60531331,
        1.04524314,  0.7637124 ,  0.0458202 ,  0.63044718,  1.02387605,
        0.6441506 ,  0.15202829,  0.06836975,  0.12113314,  0.19732339,
        0.78233429,  0.37779053,  0.75745862,  0.29348834, -0.08458378,
       -0.07173513,  0.73006681,  0.38585976,  0.09324021, -0.02912595,
       -0.10779946,  0.22953974,  0.24480956])

==> I wanna know why the probability here is not between 0 and 1 please

Legna
  • 53
  • 5

1 Answers1

3

I would suggest you use the LGBMClassifier wrapper as a more user-friendly version lightgbm.train for classification problems which provides the predict_proba method of this class to avoid such issues:

# Initialize the LGBMClassifier
lgbm_classifier = LGBMClassifier()

Fit the model


lgbm_classifier.fit(X_train, y_train)


Make predictions


y_pred = lgbm_classifier.predict_proba(X_valid)


Calculate F1 score


f1 = f1_score(y_valid, y_pred[:,1] > 0.5)
print(f"F1 Score: {f1}")

You can read more about this here.

mornington
  • 46
  • 2