books-page-count-predictor

Book Page Count Predictor

Model Details

• Model Type: AutoGluon Tabular Predictor (Ensemble)
• Task: Regression (Page Count Prediction)
• Framework: AutoGluon 1.4.0
• Training Data: Augmented book dimensions dataset
• Input Features: Height, Width, Depth, Genre
• Output: Predicted page count (integer)

Performance Metrics

• MAE: ~15–25 pages (varies by test set)
• RMSE: ~20–35 pages
• R² Score: 0.85–0.95

Intended Use

• Predict book page count from physical dimensions
• Publishing industry applications
• Library cataloging assistance
• Book manufacturing planning

Limitations

• Trained on limited genre categories (0, 1)
• Assumes standard book formats
• May not generalize to unusual book types (e.g., children’s board books, art books)

Training Details

• Dataset: its-zion-18/Books-tabular-dataset
• Train/Test Split: 80/20
• Training Time: ~5 minutes (300s limit)
• AutoGluon Preset: best_quality

Usage Example

Load Native AutoGluon Predictor

from autogluon.tabular import TabularPredictor

# Load the model (unzip first if using the .zip version)
predictor = TabularPredictor.load('path_to_model')

# Make prediction
sample = {'Height': 9.0, 'Width': 6.0, 'Depth': 1.0, 'Genre': 0}
prediction = predictor.predict(sample)
print("Predicted page count:", int(prediction))