Enhancing Babies’ Sleep Schedule Prediction through Machine Learning

Abstract

In recent years, there has been a growing interest in improving sleep quality and understanding sleep patterns. This thesis will focus on enhancing sleeping schedules for babies through machine learning. Establishing a consistent bedtime routine at a young age is crucial, as it offers numerous health benefits and can prevent sleep-related issues later in life. Despite this, many parents still find it challenging to manage their babies’ sleep schedules effectively. This master’s thesis explores the integration of machine learning algorithms into babies’ sleep schedule predictions to provide more accurate and personalized recommendations. It focuses on the integration of advanced data analytics and processing techniques to improve sleep forecasts. Recognizing the importance of data cleanliness and processing efficiency, the research delves into different steps for preparing and analyzing the dataset on sleep and baby tracking information. With a strong focus also on feature analysis, it later dives into various machine learning models and assesses their effectiveness and performance. The regression task with machine learning models includes K-Nearest Neighbors (KNN), XGBoost, Random Forests (RF), Long Short-Term Memory networks (LSTM), and Recurrent Neural Networks (RNN). The project offers a methodical approach that includes background information, relevant literature, dataset specifics, suggested techniques, findings, and conclusions. The results demonstrate a clear potential to improve the current sleep schedules with machine learning to achieve the desired goals, with performance metrics showing proximity to baseline expectations. This thesis contributes to the field by advancing the methodology of baby sleep tracking, ultimately aiming to enhance the well-being of infants and ease the challenges faced by parents in managing their babies’ sleep routines.