This wide-and-deep model combines the memorization capabilities of a linear model with the generalization capabilities of deep learning that can allow us to create a recommendation system that can predict a wider variety of choices for users. In this case, a 20 million entry movie rating and table data set from Kaggle was utlized to build a sample movie reccomendation system as a capstone project for my Data Mining course.
- Statistics
- Data Science
- Data Analysis (Pre-processing and Cleaning, Analysis, Visualization)
- Advanced Mathematical Logic (Linear Algebra, Vector Calculus)
- Machine Learning (Linear Models and Deep Learning Models)
- Python ML & DL Libraries: TensorFlow, MLflow, NumPy, Pandas, Scikit-learn, Keras
+Python:A high-level, versatile programming language that is widely used in data science, machine learning, and artificial intelligence due to its readability, simplicity, and large community support.
+TensorFlow: Framework for deep learning.
+MLflow: Manages machine learning workflows.
+NumPy: Supports numerical calculations.
+Pandas: Handles data manipulation and analysis.
+Scikit-learn: Machine learning algorithms and tools.
+Keras: User-friendly interface for building neural networks.
+Jupyter: an interactive computing environment, allowing users to create notebooks that integrate code, visualizations, and narrative text. I documented the process in various notebooks for ease of understanding and learning! :)
data_preparation.ipynb: The purpose of this notebook is to prepare the dataset we will use to build the wide-and-deep recommendation model.
feature_engineering.ipynb: In this notebook, we will engineer the features we will use to build the wide-and-deep collaborative filter recommender.
model_preparation.ipynb: In this notebook, we train and evaluate the wide-and-deep collaborative filtering recommender using features engineered in the prior notebook.