Statistical model

Jan 8, 2019
Hi all,

I am trying to figure out how to statistically determine the best location for a second sensor. I have a primary sensor (center of attached image) that periodically gets blocked. I plan to place a second sensor at pre-planned locations and take measurements. I expect to get time correlated data that looks like this:

Time Pos1 Pos2 Pos3 Pos4...etc (probably 80 different positions)
14:31 35 34 36 35
14:32 34 35 35 35
14:33 36 38 34 35
*Higher numbers indicate a better result

I have only one sensor. So I will move that sensor every day and take a new set of data. At the end I’d like to know which position provides the least blockage. I thought about just averaging the numbers, or counting how many measurements fall below a threshold. But I’m hoping to get a more accurate answer with statistics.

A quick explanation of the attachment:
- The green rings indicate a fixed distance from the center
- The lines are direction in degrees
- I plan to name the positions by ring then direction (i.e. c270, or g135)

As a bonus, I’d like to also plot the data. Like a heat map, where the better locations are dark-blue and the worse locations light-blue. To plot it, I need a result for each position. Maybe something like percentage of time a position exceeds/lags the primary measurement. Not really sure.

Found something called "Anova: Two-Factor Without Replication" in the Excel Data Analysis tools. Looks like it might be right, but I really have no idea.

Any help is greatly appreciated!!!stats_problem.jpg