Currently working on a machine learning algorithm that will predict a "fun factor" attributed to the local beach break 4 times a day by human employees of a local business. The human employees also estimate the "wave height" and "surface quality".
Wave height consists of values like "ankle/knee", "knee/thigh/waist", "chest", "chest/head", etc.
Surface quality consists of values like "choppy", "bumpy", "bumpy/clean faces", "clean", "glassy".
I am currently mostly concerned with wave height. The closest thing I could find an answer for was human height in this answer: https://www.quora.com/What-type-of-data-nominal-ordinal-ratio-interval-is-a-person-s-height-and-why-Note-that-it-allows-for-two-possibilities#:~:text=Height%20is%20a%20ratio%20variable,who%20is%203%20feet%20tall.
That implies height is ratio data.
But human height and a human-observed/estimated wave height have some inherent differences... the most obvious to me being human error due to the imperfect nature of our eyes and the motion/short life-span of the peak of a wave vs. measuring a human with a tape measure.
The employees are given a free form text to fill in when reporting, so I've taken the liberty of addressing things like typos that are clearly errors, but I'm left with high cardinality still, about 50 values for wave height, and the first step in dealing with that is understanding what kind of data I'm working with...
My plan is to line up NOAA data with the human-observed wave height and surface quality reports, and then predict the fun factor.
Can anyone help with this? If you'd like to provide any information regarding surface quality, that would also be appreciated...