The Inside Track

Because of some problems with Sleep Stats grabbing bad data (from vacation days, weekends off, and otherwise incomplete days), I made some pretty significant changes to the way Sleep data is handled. You can now tell TT to ignore any date when calculating averages and stats.

There is a new button on your Sleep Summary page that appears under the Sleep Journal area. It says “Ignore this day”. (Hold your cursor over it to get a longer description.) Clicking this button tells TT to ignore this day when crunching your data.

After you have marked a day as ‘ignore’ the button changes to red, and says “Include this day”. If you decide the data on that day isn’t all that bad, you can easily choose to include the day in your averages again.

Days marked as ‘ignore’ will show up with a red description on your ‘Full chart’ page, so you can use your full chart to see how many days you are ‘ignoring’.

Here are a couple of tips:

• If a day is already listed as “no data”, it’s not necessary to ‘ignore’ the day. TT already excludes days with no data.
• Look at your stats page to see if bad data is being including in your averages. Clicking “view” from the stats page will take you directly to the day in question and you can easily mark it with ‘ignore’ if necessary.
• Excluding days is a Sleep only feature for right now. Eventually the other telemetry will catch up, but I really have to get the site launched first

[Update]
I changed the way ‘ignored’ days show up on the chart. The colors for ignored days are now grayed back so you can visually ignore them as well computationally.

TT now separates out naptime from the total sleep-time. You can see this on the sleep summary and the home page.

Here’s how the nap total is calculated:
Any sleep entry where the ‘Fell asleep‘ time is between 8am and 6pm is counted as a nap. This is the inverse of how the overnight sleep-time is calculated (6pm-8am).

This may not be the perfect solution. The advantage to defining a specific time span is that you can accurately compare data to the group average (forthcoming). The disadvantage is that there may occasions when your data steps outside the defined area. For example, if your child gets up at 5am and takes a morning nap from 7:30 to 8:30, that nap wouldn’t be counted toward the daily nap total (it would actually get added to the overnight sleep total).

Even if each user were able to define their own day and night times (for example, day: 6am-5:40pm or night: 7:15pm-8:30am) it is likely that you would run into exceptions from time to time that would cause the calculations to be off.

Eventually I plan to develop a smart algorithm that will distinguish naps from overnight sleep. In the meantime, I think this solution will work — especially as your child gets older and on a better nap schedule.

One note: Please don’t change nap or sleep times to force them to be counted as either naps or overnight. It might make your nap or overnight totals ‘seem’ a little more accurate for the day, but it’s actually damaging the integrity of your data over the long run. The best thing to do is to keep as accurate records as you can, and wait for TT to get smarter thanks!

The Sleep Probability chart uses a gray scale to display the probability of your child being asleep at a certain time of day for the selected dates. Areas of high contrast (black and white) mean your child is on a predictable schedule. Areas of low contrast (light, middle and dark gray) mean a less regular schedule. A uniformly gray chart would mean a completely random sleep schedule.

Sleep Probability Chart for a Newborn (birth to 1 month)

6 Months Old

12 Months Old

Essentially, this chart shows your child’s sleep schedule by compressing daily sleep charts into one image. This kind of compression is extremely good at showing how patterns form and evolve over time. For example, it can help you identify the average bedtime or realize that your child is transitioning from 3 naps/day to 2 naps/day.

The number of gray scale values depends on the number of days in your chart. If you have two days in your chart, then there will be three colors: black, 50% gray and white. The reason is that the different possibilities for being asleep at a given time for the two days would be:
2/2 (100% – asleep both days at a given time),
1/2 (50% – asleep one day, not the other at a given time) and
0/2 (0% – asleep neither day at a given time)

The more days in your sample, the more possible gray scale values. So if you have a months worth of data (31 days max), there will be 32 shades of gray:
(0/31 through 31/31). This would be approximately 0% black, 3% black, 6% black, 10% black, etc…

(What? You’re not tracking yet? It’s easy to create these cool charts for your baby. Discover Trixie Tracker and sign up for the Free Trial today.)

Related links: this chart was first seen on the Trixie Update in March 2004.