This very good!
OK, so your "1 year ROI" calculation is more like "number of bitcoins mined in a year" given different average difficulty increases every 2 weeks (and your sheet 2 shows historical data for the difficulty increases).
I think "number of bitcoins mined in a year" would be a better name for this calculation.
You then need to subtract the yearly electricity cost and the original cost of the miner from the "number of bitcoins mined in a year" to get the "Profit" which will often be less than zero (i.e. a loss)

The ROI will then be "Profit" / "original cost of the miner" and will be a percentage.
Now that I downloaded your spreadsheet to Excel, I realized your "1 year ROI" is actually "Profit" as I described above. Dividing this by the original price to get the ROI is very instructive.
Basically, if you think difficulty will rise slowly, KNC miners look best, but if you think difficulty will rise faster, evilscoop's unit looks best (assuming the Lead time (maybe should be called "Delivery date") you give). This makes sense, obviously.
BTW, from the Excel file I downloaded, it seems you could make much greater use of formulas (VLOOKUP etc) to ease updating of this spreadsheet. If you want help with this, drop me a PM. (I'm familiar with Excel, but not so much, Google spreadsheet).