Step rate while walking/hiking has irregular outliers

I am just not seeing what you are seeing. Mine are generally very rock solid (am running latest beta - but was same with previous official live version).

Based on the much higher rate in your graph, I'm going to assume that this is a running activity. It is possible that it is more accurate here. But you also have at least one outlier which is also exactly twice of the real cadence. 

For me, this occurs with walking activities. The sequence of steps is slower. And especially during hikes, the individual steps are not 100% the same, in contrast to running, because I'm not walking on flat paths here, but through the terrain. 

The behavior I see here in my evaluations hasn't really changed over the last versions either, so in my eyes it's more of a general problem with the step detection of the clock. 

I'm not seeing what you are seeing either.  Here's one of my recent walking activities:

The gap in the middle is when I stopped to make a pickup.

Since the recorded pace is picked up by the movement of your wrist, not your feet, I would suggest that the irregularities you are seeing are due to erratic wrist movements you may be experiencing due to the terrain you are traversing.

HTH

Thank you for your answer, but what exactly is the solution? That I tie the watch around my leg?

Sorry, but the pattern is clear. Under certain circumstances, which of course have something to do with the fact that the movement is measured at the wrist, the detected step frequency jumps to an integer multiple without any associated change in speed, which would argue against a spontaneous sprint here.

In this respect, this is a matter of the software, and Garmin should refine it a little so that such implausible outliers are filtered out.

I had a similar problem with a cadence sensor for a bicycle that gave incorrect values. Strava simply filters out these outliers, with Garmin they stay in. 

No walking activity. 125spm is pretty usual for me.  Running cadence I like to try and get over 170spm - ideally should be btwen 174-185 - ideally around 180spm.

I don't consider the odd outlier and issue either - could be numerous reasons - perhaps raised my arm very quickly  - perhaps lead was in left hand and dog jerked it - can be a huge variety of reasons.

Have you also considered that maybe those anomolies can be due to looking at your watch, or perhaps adjusting arm to account for terrain. When I hike especially over rougher terrain my cadence is not nearly as "static" and there are much more drops and bumps to it.

See a hike I recently did (was on an official version): 

And I think its only normal considering that arm movement will not be regular.

Solution - I don't think there is one unless prepared to pay more attention to arm movements; but I don't know why you are so worried about it as its just walking. Its good to know what your cadence is but other than that what do you plan to do with it?

Its not like running where cadence is more important as it gives info about ground contact time.

I completely agree with you that there can be various reasons why there are these outliers.

However, the watch that explicitly has a "hiking" profile should also tolerate corresponding movement patterns. Because if it shows up exactly as it does here, it can be easily filtered out. The apparently recognised stride frequency is simply not plausible for a person of my height and average stride length -- especially when the speed is taken into account.

If I were to use an aerobic sport profile here (or any other sport where you take quick steps or jumps), then such a pattern would be expected. But not when walking or hiking. There, the expected step frequency lies in a range that results from the step length and distance covered.

On a short hike yesterday, the same thing happens again. The frequency jumps between 1x and 2x the actual cadence.