Back in February in my first post of using Apple Watch to track swim, I wrote about the metrics available and usage in the Apple Watch, a handful of data quality challenges, and my plan of using data captured to improve performance. After logging another 22 workouts since Mar 2017, I would like to give an update, and also share a few more new interesting lessons learned.
watchOS 4 is solid
Slightly disappointed by my “outdated” Apple Watch series 2 (Series 3 was announced) less than 1 year, I am surprised to see that Apple made the following improvements in its latest software (watchOS 4):
- The new Heart Rate(HR) chart: This is a typical example of using visualization to drive actions. In the legacy watchOS 3, there is only one “Avg HR” for measuring intensity. I had doubts about the number in the past, but never bother to dig deeper for the data points breakdown. They were located in another app (Health) and required multiple clicks.
When looking at this new HR chart after my 1st workout(above chart), I immediately noticed the problem. There is no way my heart was only ranging around 92 – I even have average 100 when walking with the baby stroller! The only plausible reason is the data is not captured properly – the loose watch strap is likely causing it, especially underneath the water. After making the adjustment in the following week, the 2nd chart looks much more reasonable.
Takeaway: Critical details might be hidden inside the “average” metrics. A proper visualization could maximize the signal could lead to favorable actions.
- Auto Sets: With this feature, you can finally see the breakdown of each different swim style, for example 675m of freestyle and another 175m of breaststroke. It should also address the data loss issue when I initially posted 8 months ago when changing the gesture. But I have not personally tested it yet – I “borrowed” the screenshots from others.
Takeaway: Even you don’t get it right or perfect the 1st time, it’s okay. As long as you acknowledge it and commit to fix it or make it better. In my case, switching to all freestyle swimming to accommodate the data collection flaw isn’t a bad choice – It has clearly made me a better swimmer.
My two decisions made – not data-driven.
In my previous post, I also shared two of my pending decisions:
One decision approaching is whether I shall switch my swimming location from the university gym (which usually offers 50m long pool) to the local YMCA(only 25m pool available). I used to prefer the 50m pool a lot more. It means I only need to make half of the turns, comparing a 25m pool. I don’t like to make turns, as I was worried losing momentum and speed. With the new watch, I noticed I actually swim faster in a 25m pool than 50m, however in a small sample size comparison.
Ironically, I opted in for both gyms. The university pool has treated me well in the last 3 years, and the local Y is not optional for a family reason. At the end of the day, I realized I like swimming much more now, particularly with the ability to analyze my own performance. I made the upfront financial investment, hoping to increase frequency from 1 time/wk to 1.5-2 times/wk. That also means more data points will be available for me to analyze and play around.
The other decision I am pondering right now; is how shall I do with swimming next (how can I improve in the future). There are 2 options – I shall either improve my intensity (swim faster in every 1500m workout) or improve my durability (push my workout to a longer distance – for example, 2000m). I will be very curious to know, how can I use data collected by my apple watch to make this decision.
Maybe next step is to do a little bit of A/B testing – swim faster for a few times, and then swim longer for a few times, then compare the result – which option can burn my calories (my expectation will be swim longer, but by how much?), which option will help me to reach my goal – which is to keep & improve my fitness level. I don’t know if the metrics provided by the watch will help, but at least it worth a try.
This decision is made as well – I have decided to swim faster in a shorter workout – now I do 800m-1200m more often than 1500m. At my current life stage, I simply don’t have the 2hrs time to complete a 1800m or 2000m workout, adding the shower + commute time.
Takeaway: As an analyst, I am always striving to make data-driven decisions. In reality, there are many other factors fell out of the scope of data. When restricting myself as the role of “analyst”, it is easy to run a bunch of A/B test and tell the decision maker – You should do the longer workout because it burned more calories. However when I take my “analyst” hat off and put my “decision maker” hat on, the “warm, family commitment” time limitation factor has clearly outweighed the “cold hard” calories.
Next step: advanced analytics tool
Although it is hard work spending time analyzing the data and organizing my thoughts in a long-form written format, it is actually quite rewarding and fun. There are so many benefits of keep swimming and I won’t waste my time naming here, but it is a lonely sport. It is kind of boring. This is my way of getting motivation.
After laying down my options, I am planning for the following two as my goals for next 6-8 month.
- Use an advanced tool to analyze my own swim data: Inspired by Ryan Praskievicz’s post on using R to analyze Health data , I decided to use Python to do the same thing and write the 3rd post to share what I have learned, targeting in the first half of 2018.
- Improve speed within the same targeted distance: My average speed per 100m has also improved from 3’10” in early 2017, to 2’45” in late 2017. My next step is to stabilize my speed in the 2’30” mark for the 1000m workout. I am curious to see when this can happen. My best estimate now is after 10 workouts. We have 8 weeks left in 2017 as of today.