Trends - Advice?

With the latest update, I don’t seem to have the written advice anymore - the “In order to maximize deep sleep percentage, your fall asleep hour should be . . . .” etc. I only see charts and graphs. Was this advice section removed for some reason?

Hello Ruby, big sorry… yes we removed it due to bigger rewrite of the stats engine… Also we have long though about either redesigning it or removing it… sorry for the issues if you were used to this advice we will try to reintroduce it in some better shape hopefully…

1 Like

Thank you for the response! I definitely found the advice helpful, but I understand apps change over time. I do hope it is reintroduced in a future release!

2 Likes

Many thanks we will look into it… all the best!

1 Like

Hi Petr, you requested for me to follow up this topic at the forum.

I would like to clarify further from your email that, even if the Trend | Advice comments by the app do not make a return, I found the previous version of that data chart at Trend | Advice to be much more helpful for optimizing sleep. Because I could see how the app created those comments by noting the points of the most optimal upward slope in the graph.

Using the new simpler graphs, it is very difficult to see the graphs well enough on a small screen to figure out most optimal sleep, unless the comments make a return to explain what those simpler graphs are trying to say.

Another reason I would like to see the comments on the Trend back is that it’s a really nice unique feature of your app that, as far as I’m aware, no other sleep app has.

I have updated to the BETA version now. You wanted to know if I’d received a fix for this issue that way - I have not seen a change to this feature for me yet.

Hello folks, I have a good news, I’m returning the advice regression statistics back into a BETA today… I hopefully fixed the main issues with overfitting of the models, so now the try to use the simples model possible - usually a linear model and only if the RMS statistics is significantly better we take a more complex model like polynomial, but we never use a polynomial model with degree higher than 2…

Also I have added another sleep score measure to the statistics which IMHO could be very helpful to target - the efficiency measure… so you can see not only which fall asleep hour and duration targets best deep sleep % but also which cases you to be less awake…

Hope that help… here is a little screenshot from the new charts…

This is now released into BETA… in addition we have improved outlier filtering to form more consistent regression models, now there are separate criteria for each axis…

Please make sure you are in the BETA channel… this is coming in version 20210423… many thanks for all your help and feedback on this…

Petr

1 Like