In this work we exploited an algorithm, already present in the literature, and based on the notion of signal permutation entropy, to analyze a very long time series of sleep data from a single subject. The aim of the work is to explore methods for personalizing alerts related to sleep anomalies, and recommendations for improving sleep quality. As a matter of fact, sleep duration and sleep quality may differently affect daily performance of different people, as well as daily activities may differently affect sleeping during the night. Data have been collected from a Fitbit Alta HR activity tracker worn by the subject for about three years. Results show that personalized inferences may be very different from the generic (population-based) ones, and that correlations found may suggest subject-specific life-style modifications useful to improve sleep quality.