Every month Switch has led on Amazon it has won NPD. Switch did not win November on Amazon despite this so called un predicted ratio. On top of that November is a largely retail driven month not an online one.
Of course these numbers can be interpreted in different ways but to me Amazon over-represents the Switch. It also seems that the offset of Switch sales leaning towards Amazon increase over the months.
Of course, Amazon is not a precise measure. There are a number of factors which may have an impact on the ratio. These can be predicted as well, however, and I believe that Amazon remains very much relevant data to analyze.
For example, November, despite it being close on Amazon, the Switch was far behind - because Black Friday deals predictably resulted in higher retail store sales for PS4 and Xbox.
I think, over time, the value of the data we receive from Amazon will, in fact, *increase* - early switch sales fairly directly described how many systems a respective retailer got from Nintendo (if Walmart is all out, more people will buy the system from Amazon). However, now that the system is widely available, demand should be largest driving factor to new numbers we obtain.