umap VS Openstreetmap

UMAP GitHub repository: https://github.com/lmcinnes/umap

Uniform Manifold Approximation and Projection (UMAP) is a well-known dimensionality reduction method along with t-SNE.

A number of excellent topic modeling libraries exist in Python today. BERTopic and Top2Vec are two of the most popular. Both use sentence-transformers to encode data into vectors, UMAP for dimensionality reduction and HDBSCAN to cluster nodes.

As with anything, it depends on the problem. But T-SNE and UMAP are often good.

UMAP distances.py: umap/distances.py at master · lmcinnes/umap (github.com)

I used for this project Flask and OpenAI CLIP. For the vector search I used approximate nearest neighbors provided by spotify/annoy. I used Flask-SQLAlchemy with GeoAlchemy2 to query GPS coordinates. The embedding was done using UMAP.

side note: instead of t-SNE consider UMAP - provides better results (and it's much faster) https://github.com/lmcinnes/umap

Sounds like a job for UMAP https://github.com/lmcinnes/umap ?

I am a fairly experienced python developer/researcher (about 10 years), and have found a bug that breaks all of my intuitions. I am messing with the [UMAP](https://github.com/lmcinnes/umap) repository and trying to add the option to disable some additional features. I've stripped everything from it but have a [quick test that will run my UMAP version and compare the outputs with what the original gave](https://github.com/Andrew-Draganov/probabilistic_dim_reduction/blob/master/umap/nndescent_umap_test.py). Managing my random seeds, same inputs, all that.

I'm currently reading through a project on github, https://github.com/lmcinnes/umap, and in `umap/umap_.py` at line 2287, they have this:

I wouldn't compare osm-website and osm-carto at all. The commit logs are very different.

openstreetmap-website: https://github.com/openstreetmap/openstreetmap-website/commi... . Numerous commits most days.

openstreetmap-carto: https://github.com/gravitystorm/openstreetmap-carto/commits . So far in 2024; one small regression fixed, one niche bit of tagging added to an existing style, some largely pointless code style tidying. That's it.

Contributing is simple:

1. When you see a trail or any other feature that doesn't appear on the map, take a picture.

2. When you get home, visit https://www.openstreetmap.org and start drawing.

The website has satellite images overlayed wirh map data, so it's easy to see what you are doing.

You can look at your pictures to remind yourself of what was missing.

If you have recorded your ride,you can also upload your GPX trace to OpenStreetMap to make it easier to trace features that don't show up clearly on satellite images.

What impressed me was that it looks like openstreetmap shows the bridge as down already.

https://www.openstreetmap.org/#map=16/39.2144/-76.5279

Through the Fast Forward program, we give free services and support to open source projects and the nonprofits that support them. We support many of the world’s top programming languages (like Python, Rust, Ruby, and the wonderful Scratch), foundational technologies (cURL, the Linux kernel, Kubernetes, OpenStreetMap), and projects that make the internet better and more fun for everyone (Inkscape, Mastodon, Electronic Frontier Foundation, Terms of Service; Didn’t Read).

Way overdue. OpenStreetMap's website at openstreetmap.org is its calling card, and for the past few years the default style shown (called Carto) has all but stagnated in development. Accepted features like highway=busway (introduced three years ago) are not rendered there because the maintainers can no longer be bothered, or dislike the tag personally despite broad community backing.

What worries me for this new effort is that Paul Norman is one of the two remaining Carto sometimes-active maintainers who refuse to merge contributed PRs or even provide alternative minimal support for features like highway=busway, leading to awkward gaps on the baseline map shown on openstreetmap.org.

I would love to be surprised in a positive way about this new effort, but I'm not holding my hopes up. Thankfully OpenStreetMap can be thoroughly useful in apps like OsmAnd and OrganicMaps, and the tile-based Tracestrack Topo layer on openstreetmap.org is getting quite decent:

https://www.openstreetmap.org/#layers=P

https://www.hotosm.org/tools-and-data runs software that's used for example after an earthquake. The tasking manager specifically is a reactjs app plus postgresql with plenty of open issues. HOTOSM has full-time staff, I'm not sure if the developers are full-time, but it's more organized than a volunteer project.

https://github.com/openstreetmap/openstreetmap-website/ is Ruby on Rails (easy installable with a docker setup). The maintainers have trouble even reviewing incoming PRs so an experienced person who can triage, test, review is currently needed.

If you're in the US then https://en.wikipedia.org/wiki/Code_for_America might be worth having a look at.

https://summerofcode.withgoogle.com/ will soon annouce vetted organizations who do open source. (last year https://summerofcode.withgoogle.com/programs/2023/organizati...). Project are paid, the process is long though, all summer. https://developers.google.com/open-source/gsoc/timeline

> 9450 S. W. Barns Rd

Portlandians: Are Barns Rd and Barnes Rd the same thing? Looks like a nice spot if so: https://www.openstreetmap.org/#map=18/45.50901/-122.77468

That building is now a SFX agency: https://hellohinge.com/ (No relation to the dating app)

Also curious if the TEKsystems employment agency next door took its name from Tektronix.

Yes, https://www.openstreetmap.org has quite inconsistent detail as it relies on people mapping stuff.

And help is welcome, anyone can join and help with mapping!

It's shown on OpenStreetMap, but not as a street: https://www.openstreetmap.org/#map=16/51.3682/1.0330