The book’s rating predictor is a model-based project capable of predicting the average rating of a book based on variables such as the author name, the average rating count, the publisher, the language in which the book is written, the number of pages, and the publication date.
The different machine learning models are based on the Goodreads database which can be found at the following address: https://sites.google.com/eng.ucsd.edu/ucsdbookgraph/home. The database contains 11127 books and 11 features for each book. The original data contains 4 rows with extra columns; this is due to an extra separator (;) between the authors ‘names. the cleaned dataset has been uploaded to our GitHub here https://github.com/hassnabdl/PythonLabsProjectS23.git . To explore this project please download the dataset (books.csv) and the python notebook
Insert information about a book. Get a prediction of his average rating. The project is submitted as a Jupyter Notebook and includes data preprocessing, exploratory analysis, feature engineering, model training, and evaluation. Based on the accuracy of each model, the best model is the Random Forest model.
To train our machine learning model, we needed to preprocess our database. The model cannot work with categorical variables. This is the reason why these variables have been encoded (see target, label, and frequency encoding).
If you need support you can follow the contributors on GitHub using their usernames: Arnelle Nguefang Tchouamo : Arnellou;
Hassna Boudali : hassnabdl;
Yilun Du: Jed-NOBODY;
Djihene Beladjine : DjiheneBA;
Kaoutar Zeroual : Kaoutar-ZER