Where to find advice
After gathering at least 2 weeks of data Sleep as Android starts to provide your with advice on how to improve your sleep and what are the current trends in the data.
Advice may be Positive, Negative or Neutral. For example having a gain in sleep length in case you are under the recommended values is considered positive. But consuming more alcohol is considered negative. Having significantly different levels than the population average is considered neutral. You get +1 for each positive and -1 for each negative advice resulting in an overall sleep score. Having an overall positive sleep score means you are doing well and there is more positive than negative trends in your data.
We keep track of the most important sleep parameters and their development. This advice compares your last 7 days development with your last 30 days development. Such advice takes the form of e.g.:
Negative: Your “deep sleep” is down by “10%” recently.
Another kind of advice keeps an eye on your sleep parameters to be in the recommended ranges based on larger scientific studies. For example for adults it is recommended to sleep between 6 – 8 hours. If you sleep falls under 6 hours we display:
Negative: Your “sleep length” is under “6 hours”.
Tags are another great source of advice. Example:
Positive: You are getting more #sport recently.
Negative: Your are getting more #caffeine recently.
Advice in Sleep as Android further compares how you are doing in comparison to others using the app. This data is based on anonymized aggregates gathered in the SleepCloud online service. Example:
Neutral: Others are getting “30 minutes” less snoring.
Maximize your parameters
Another source of advice is the light bulb marked section in charts. It helps you to find the right sleep parameters in terms of fall asleep hour and sleep length to maximize your deep sleep or subjective ratings. In other words it tells you when to go to bed or how long to sleep in order to get most deep sleep % or in order to get sleep which you tend to rate high.
See more about polynomial regression – the statistical method behind this in How it works.