LS-W4-Mini-RF_Addiction_Impact

Model Summary

This is a Random Forest Classifier trained to predict whether social media use affects a student's academic performance. The model is based on the "Social Media Addiction vs. Relationships" dataset from Kaggle, which contains survey responses from students aged 16 to 25.

Usage

The model is packaged within a scikit-learn pipeline and can be easily loaded and used within any Python environment. It expects a pandas DataFrame with the same column structure as the original training data.

import joblib
import pandas as pd

# Load the model
model = joblib.load('LS-W4-Mini-RF_Addiction_Impact.joblib')

# Example of new data to predict on
new_data = pd.DataFrame({
    'Gender': ['Female'],
    'Academic_Level': ['Undergraduate'],
    'Most_Used_Platform': ['Instagram'],
    'Relationship_Status': ['Single'],
    'Age': [20],
    'Avg_Daily_Usage_Hours': [5.0],
    'Sleep_Hours_Per_Night': [6],
    'Mental_Health_Score': [7],
    'Addicted_Score': [8],
    'Conflicts_Over_Social_Media': [0]
})

# Make a prediction
prediction = model.predict(new_data)
print("Prediction (1 = Yes, 0 = No):", prediction)

Training Data

The model was trained on the public dataset Social Media Addiction vs. Relationships. The dataset consists of 705 records and 13 features with survey responses. The training data and the model file are available within the repository.

Model Details

• Model Type: scikit-learn RandomForestClassifier
• Pipeline Structure: The pipeline includes a ColumnTransformer for one-hot encoding categorical features and the RandomForestClassifier itself.
• Key Hyperparameters: n_estimators=100, random_state=42.

Performance

The model's performance was evaluated on a held-out test set from the original dataset.

• Accuracy: 0.93

Limitations and Ethical Considerations

• Not a Diagnostic Tool: This model should be used as a statistical tool for trend analysis and should not be used for clinical or psychological diagnosis of addiction. The data is based on self-reported survey responses.
• Generalizability: The model was trained on a specific sample of students and may not generalize well to other populations, age groups, or time periods.
• Data Bias: The model's predictions reflect the biases present in the original dataset. The results should be interpreted with caution.