Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Do I need 'locators' section in config? #906

Open
SurreyMichael opened this issue Dec 11, 2024 · 2 comments
Open

Do I need 'locators' section in config? #906

SurreyMichael opened this issue Dec 11, 2024 · 2 comments

Comments

@SurreyMichael
Copy link

I adapted the sample config.yaml file that installed with companion but couldn't figure out the locators section (see below), which isn't mentioned in the guidance docs. Are these used/essential? Should I enable one but not both? Any advice would be much appreciated.

ESPCompanion is working in my setup but I'm still tinkering to see if I can improve accuracy and response.

`locators:
nadaraya_watson:
enabled: true
floors: ["first", "second", "outside"]
bandwidth: 0.5
kernel: gaussian

nealder_mead:
enabled: false
floors: ["first", "second", "outside"]
weighting:
algorithm: gaussian
props:
sigma: 0.10

nearest_node:
enabled: true
max_distance: 10`

@dxmnkd316
Copy link

dxmnkd316 commented Dec 13, 2024

I've been playing around with both. I have nadarayawatson enabled and find it's pretty good for performance. I leave the other two disabled for now.

But Gaussian was the best prior to NW for me with a moderate sigma of 0.333333. I'm still not entirely sure how the NW works. I did some reading on the mathematical method on Wikipedia IIRC but didnt have time to really dig into the nitty gritty. I suspect the bandwidth is sort of a fuzziness around the beacon distance to the node. So it's not trying to find a perfect location with multiple nodes that just can't actually be solved. But I could be entirely wrong on that.

Edit: here's the article I was reading:
https://en.m.wikipedia.org/wiki/Kernel_density_estimation

I was wrong. It's not fuzziness from a single datapoint. It seems like it's more about multiple points and the density of the signals. So it should give a better result than just Gaussian weighting.

Edit 2: an even better discussion of the general approach.
https://en.m.wikipedia.org/wiki/Multivariate_kernel_density_estimation

@SurreyMichael
Copy link
Author

many thanks - really good to hear what you have tried as I couldn't find anything about this in connection to this add-on. I'll enjoy reading the links

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants