This model is a Random Forest Classifier designed to predict whether social media usage has an impact on a student's academic performance. The model was trained using the "Social Media Addiction vs. Relationships" dataset, which is a collection of survey responses from students aged 16 to 25. The model is built as a scikit-learn pipeline, which includes data preprocessing steps and the classifier itself, making it easy to load and use in a Python environment.
The model's primary use case is to provide a statistical tool for trend analysis regarding the relationship between social media use and academic performance in a specific student demographic. It is suitable for academic research and data exploration, but should not be used as a clinical diagnostic tool.
The model was trained on the Social Media Addiction vs. Relationships dataset from Kaggle. The dataset contains 705 records and 13 features, all based on self-reported survey data. The pipeline's structure includes a ColumnTransformer for one-hot encoding categorical features before they are passed to the classifier.
Key hyperparameters used for the RandomForestClassifier were:
n_estimators: 100
random_state: 42
The model's performance was evaluated on a held-out test set from the original dataset, achieving an accuracy of 0.93.
It is crucial to understand the limitations of this model for responsible use:
Not a Diagnostic Tool: The model should only be used for statistical analysis and should not be used to provide clinical or psychological diagnoses of addiction. The data it was trained on is based on self-reported surveys, which may be subjective.
Limited Generalizability: The model's predictions are based on a specific sample of students. It may not be generalizable to other populations, age groups, or different cultural contexts.
Data Bias: The model's predictions reflect any inherent biases present in the original training data. The results should be interpreted with a critical and cautious approach.
The model can be loaded and used in a Python environment using the joblib library, as shown in the example below.