This is cool but I don't think it fits my use case..
Seems like I have to pick criteria that have exact venues.. I want to pick abstract things like "walking distance from grocery", "biking distance from climbing gym" "1 hour drive to national park"
To echo what loxias said, it is possible to make queries like this on the heatmap. You can use the "Search Nearby places" button - this takes in more general queries (like cafe, gym, walmart, etc) and gives you back a bunch of venues that fit that search.
I don't really understand what the search nearby places button is doing. I think the solution would be to allow some sort of OR operator (which seems straightforward), or conditions the act on generic queries (which seems more difficult). For example say you live in NYC and want some nice green space near you. Choosing a single park is too narrow and choosing all parks is too broad. So you should be able to say e.g. "Central Park OR Prospect Park OR Brooklyn Bridge Park OR Fort Greene Park", or you should be able to say "Park, > 10 acres, 4+ star google rating, has tennis court, has bike path"... the point is that not all parks, grocery stores, coffee shops, etc are equal; I need to be able to qualify them somehow.
The second one (acrage, stars, etc) is harder, you're right.
The "Search nearby places" is really meant to be a convenience feature to fill your OR clauses. It works better for certain types of things. Like, sure, all parks is too broad because not all parks are equal. But use it for something like all Targets (the shopping chain) or something, and its more useful, since those are, more or less, all equal.
I'm still thinking of other convenience features for places that have more nuance, like parks.
I found it was surprisingly easy to "populate with every instance of a given type". I've made a few maps based on grocery and transit where I did it by adding all >100 stores & stops.
"Total count of non-national-chain restaurants within a 1km walk" pretty much determines my criterion. Everything else is a once a week thing or less. The homes themselves are just a bunch of rooms, and I'm flexible. Quality of life is defined by my stomach.
I currently commute by train _and_ bike, but this only lets me filter by one or another, unfortunately.
This seems to be a common problem with navigation systems in general. It's easy to get walking+transit directions, but nigh impossible to get bike+transit, even though all the buses and trains near me let me bring a bike onboard.
Yeah the "mixed" navigation option sometimes provides it, but with many caveats:
- it never suggests bike+bus, only bike+train
- it will only ever try to put the biking on one end of the train ride, never both. I guess it assumes I'd be parking the bike at the station, not bringing it on board
- you can't actually "start" the navigation for some reason. It will just show the route overview
I have seen it schedule the bike at both ends. But it does seem like it assumes the bike is an 'option' like all the other forms of transport, which makes sense
for bikes hired by the minute, but not if you're taking your own with you - you need all segments to be bike-friendly.
I put my wife's work location and my own in it and it correctly showed me that where we actually live would be a good location. When we moved to our house she worked at a different place, but I can see how this would be useful.
The "only show best matches" criterion is a little bit too aggressive in this case, though - it basically says "have you tried living in the middle of the highway"?
Also matched my place perfectly. With my work, leisure and commute requirements! Funny, I feel my requirements overfit on my habits - the area where I live was the only highlighted area with no alternative sweet spots.
This is incredible! I've been looking at new houses for a while now, trying to balance living next to a park, walking distance to elementary school and a grocery store, and within a driving distance to my wife's work. I'm having a blast playing around with it right now and scoping out the potential neighborhoods that would be best.
One feature request: when you do "Search Nearby" (e.g. I did "elementary school") it found a _lot_ of schools, like 50 or more, well out side the city I was interested in. But short of adding them all and going through one by one and deleting them, is there a way to add just the matches in the city I put in the top-right corner? Maybe adjust the search radius or something? Or in the search results preview it gives you, a button to just select the few you want to add.
edit: oh, and another feature request. I'd love to accordion/collapse my criteria. Scrolling past my 40 parks to edit the bottom criterion is tedious.
I love the idea, but the data seems a bit off. I tested it in my little city, where I'm < 14 minutes by car from the grocery. It's actually a little quicker, but I'm going by Google Maps estimates.
Setting a criteria of 15 mins by car, I'm far out in the gray. I'd have to drive a couple miles to even get in the red. It's only 6 miles away!
I was hoping 'little city' would have indicated, but I should have specified, there is never enough traffic here to move the estimates much. Speed limit is 35ish the whole way.
Yeah, different geo analysis providers have different weightings for travel time (turns, street density, etc). It can sometimes lead to inconsistencies like this, especially if they don't use GPS and proprietary data to correct things...
TomTom maps (and probably others, I just worked for TomTom so know how theirs work) combine both real time traffic information, and road speed estimates based on data from users uploading their driving traces (with start and end locations removed to anonymize them), so it has a table of the average driving speed for all roads, on a per-hour and weekday/weekend basis, plus any real time traffic information it has, which it uses for route calculation and time estimation.
I thought about using Google maps street view photos for something like that, get all the photos for a city, and then classify how much trash is visible, and also calculate how many people there are and what they're doing.
My idea, at least from living in New York, is that if there are a medium amount of people walking around, and not that many people just sitting there or leaning against a building, and there isn't a ton of trash visible, then it is probably a pretty good neighborhood.
If there is nobody, or a ton of people, or if the people aren't mobile, or if there's a lot of trash, then it is probably not a neighborhood that I would like.
You could probably also use business types for that. Like, I would not want to live in a neighborhood that has a ton of pawn shops or churches in it.
I wanted to do this with a city I don't currently live in, but every time I did searches on specific locations it went for the nearest (bad) match to my current geographic location. It'd be nice to generalize it.
I do like that you used OSM rather than Google maps.
I should probably make the UX better. When you're on the heatmap page (or the distance matrix page), look at the top right of the screen. It shows you where it's basing its searches on, and you can override that bias with any location you like.
You only need to update it once per session - all pages and components will be updated
On the UX it would be helpful to add an option to delete all places.
It added 50 grocery stores with a "nearby places" search and inhad to refresh the page to start a new search instead of spending time clicking those X's.
This is really cool! Well done. Like others, I'd love to specify generic things like distance to a "grocery store" or "gym" as part of my initial criteria. I see that I can add a long list of possible places that meet a search, so maybe I just want the UI to hide the details from me and add all those possible places for me.
I personally found the additional criteria being added to the top to be counter intuitive and I inadvertently deleted locations thinking it was the newest criterion, but it was actually my earlier ones. I think I've been trained to look/scroll to the bottom for the added element (e.g. like when adding additional Google Maps locations)
I would also love an option to mix transportation modes. For example, public transportation and biking.
Very cool! This is nice when considering moving to a new area to narrow down neighborhoods that could work. One thing that I think could be useful is to also add criteria for things I want to be far from. Some people don't like living near airports for example.
>O'Reilly Radar has the scoop on the most recent thing I've finished working on at Stamen. Interactive travel time and house price maps for London. Go play, and read what mySociety have to say, including the ones for BBC TV Centre and the Olympic Stadium site. Then come back and read this full post if you want the background info...
I'm not in the market for a new home right now but I would absolutely use this in a future search. As it is now, it helped me confirm that my house is in a great location for everything I do regularly.
It did seem to think that the closest "Bar" to me was a 19 minute drive, when in reality there are several within a 2 minute walk, however.
I've always liked the idea of combining distance criteria, especially when something is good for public transit to work, but not good in terms of walkability to anything I care about. (Some people walk for fun, I always seem to need an errand.)
I recall that Walkscore used to have something like this, and then it went away, and then it showed up on some other housing site... I was always surprised the type of feature didn't get more popular.
In terms of new features, there is a tricky problem of how to define things like "near a grocery store, the large kind, not that one tiny mini-mart". This brings in several overlapping challenges: How to get business locations and categorize them, how to allow the user to tweak that categorization or result, and how to efficiently turn a union of those the set of valid destinations into a combined region.
Ideally something like 5min bike ride to the grocery store, 15min walk from a train station, and 30min drive to my in-law’s house.
It would be really interesting to do something like “10 minute bike ride to 3 or more grocery stores”. That would help reduce instances of niche specific stores, but also provides a much more useful variety.
I’d love to be able to find places that have 2+ or 3+ grocery stores within somewhat reasonable distance, and same thing goes for restaurants. Really any restaurant.
1. User defines a "multi-location" spec, like "MyFastFood" as "Having [2] or more of [Fast Food] excluding [Taco Bell,]"
2. User defines a requirement for their heat map which references the multi-location, ex: "Within [30 minutes] to [walk] to [MyFastFood]"
3. Within the context of a particular [user's requirement] and broad [city/town/region], a 2D area/gradient can be generated, and cached for a rather considerable period given how slowly businesses open/close.
Granted, that's the ambitious version. A simpler one would be to not support "at least X", and to combine the multi-location and the distance-rules all together into a single condition.
As others have mentioned, this kinda misses a "category" filter that is not just a search. I would suggest using the osm database for this (e.g. via overpass). If the data is good (depends on the location), this is better than just a keyword search.
I also have another issue:
If I put the same thing in two different criteria (with different settings), it says the heatmap parameters are invalid.
My use case is this:
I want to have a big station reachable within 30 minutes by public transit and any light rail station within 10 walking minutes. But this big station could fall into both of them, which, to me should then be handled according to the set criteria (maybe just treat it as two different entities entirely, just with the same coordinates?)
> If I put the same thing in two different criteria (with different settings), it says the heatmap parameters are invalid.
That's odd, since the website does not differentiate places by coordinates. I think you might have been missing something else (like you clicked a "new place" button and didn't fill out that place, maybe)
I previously used Mapnificent to choose a house location, and found that many geographically closer properties were often much further in terms of time. Very useful. I like that it starts off with the map view, maybe do that in terms of “shortest time to a-ha” and to get the user into putting in the work to refine the output?
This is an excellent concept - and it may be hard to realize but it would be worth integrating "cost of living" and "languges spoken" concerns, and then to be able to find one's perfect spot on the planet.
I've been looking for a tool like this forever.
Trying it for my city (Milan, Italy), and it's pretty accurate.
Would it ever be possible to specify which type of public transportation to consider (ie: only metro, only bus, ...)?
The two-ended Min/Max sliders of the Dynamic HomeFinder and the SimCity Frob-O-Matic Dynamic Zone Filter allow you to effectively invert selections by dragging the Max end all the way to the right, and then adjusting the Min end, since the filter passes everything between Min and Max. Or you can drag the whole slider range by the middle to adjust Min and Max at once, exploring a fixed width interval of the values.
A useful feature (not implemented in that video of X11 SimCity from 1992, but it would be easy to implement now with better graphics and faster computers) is to display a histogram in each of the sliders, where the x-axis is the parameter value, and the y-axis is the number of items of that slider's value (given that all other sliders are at their current value), so you can easily spot clusters and peaks and sparse areas, and you can include or exclude them from the filter by sliding the Min and Max edges across the histogram. So as you adjust one slider, the histograms in the other sliders change to reflect the current "slice" of multi dimensional space with respect to the filter you're adjusting.
Showing a histogram for every filter parameter on each slider gives you a multi-dimensional view of the distribution density of the data, that you can tweak and explore in real time, which helps you figure out how to adjust the filters to find interesting and ignore uninteresting items, and focus in on just the items you want.
I love histogram sliders! I think they were originally developed by Chris Ahlberg in Ben Shneiderman's group, and were later commercialized in Spotfire (I have yet to see them in any other infoviz system except Panopticon - https://www.perceptualedge.com/blog/?p=965). Jeff Heer did some work a few decades to generalize into 'scented widget' (http://vis.stanford.edu/papers/scented-widgets). With systems like DuckDB it's now even easier to implement them into various visual analytic systems.
I'm not sure why others have problems, but I love it!
Once, in a similar situation, I had the thought that "somebody should build that" and never imagined that somebody actually would.
With your tool, I learned that our home is better located than I thought and that there are many places more downtown that would be worse for me and my wife (bc of highways and good public transport connections).
You could take the aggregate of "where people want to live" and then compare that with an estimate of how much floor space the city allows to actually exist. I suspect the ratio would correlate extremely well with prices & rents per sqft of floorspace. In a perfect world this would inform urban planning, though in this world I am less confident.
This is really neat. It would be cool if you exposed some parameters like walking speed. This shows that a trip I make daily in ~12 minutes is just barely within the 30 minute walking shell from my house. Also if you want to waste an insane amount of time and effort, you could integrate this with a digital elevation model to yield better biking times.
It seems nobody has asked/suggested this so I'll do it - I would love to specify time of day/traffic conditions during which to determine time taken to travel.
In the next year or two we'll be looking to move and since we work on the southern side of the city, living out to the north wouldn't work due to traffic through the city.
I've built a similar app, but for the whole US, where you can input your personal preferences and find the areas in the US that best match based on the data:
I tried adding my city's main bike path that goes around the whole city and it seems to only register it as one point instead of a bunch of access points unless I go look up every entrance and manually add it.
Would be nice to use conditions. Like, if there are multiple libraries in a city and I would like to evaluate being 15 minutes from any of them, have a condition able to be set -
15 minutes from:
Library A OR
Library B OR
Library C
I know a homeless guy in Texas whom I talk to on IRC daily. Even he has preferences. He lives under a bridge near a library that gives him Internet access.
Great idea! But unfortunately the site was too hard for me to use on mobile (iOS, Brave browser). I was unable to set the location - when I tried to set the location, the screen would randomly zoom in and I couldn’t scroll through the addresses.
This is really, really cool and thanks for making it!!
We must think at least somewhat similarly, last few times I was apartment hunting I did the same, though I never polished it up like this (more plugging numbers into a spreadsheet).
Honestly the biggest thing this does for me is validate that the data APIs must exist for what I'd really want, which is write something to make much larger and more complex "programmatic" maps -- the list of places being generated by a more complex sequence of steps for instance, and the combining function for different criteria including nonlinearities.
Curious how you're computing the walking distances, I'm guessing this is combining some off the shelf API for it with another for the points of interest? Though it would be badass if you did it from scratch starting from just OSM. ;)
See also, Hood Maps [0], a crowdsourced map of annotations of people's opinions on particular areas in a city. It's by Pieter Levels who built Nomad List, PhotoAI, etc, you might know him if you're in the indie hacker and startup space at all.
I tried something similar myself but failed on wanting to load the whole world of openstreetmap onto my server and into postgres. It worked but I never managed to get sub second postgres query times on a billion rows. Hence, making it unsuable for any user.
So my question to you. What's your server and data setup? Do you even have your own data? I'm very curious on what is actually needed to make it work anywhere.
I’ve been looking for something like this. On my phone it’s rather unclear how to interpret the heat map though. There’s some kind of broken tooltip. What do the colors represent? It looks like it would be the distance to the closest point rather than to all the points?
How do you compute the heat map? Thousands of API requests from various points?
It reminds me of the "Dynamic HomeFinder", developed by Christopher Williamson and Ben Shneiderman at the University of Maryland Human Computer Interaction Lab in the late 80's, which they published a paper about in SIGIR '92: "The Dynamic HomeFinder: Evaluating Dynamic Queries in a Real-Estate Information Exploration System".
>Abstract: We designed, implemented, and evaluated a new concept for visualizing and searching databases utilizing direct manipulation called dynamic queries. Dynamic queries allow users to formulate queries by adjusting graphical widgets, such as sliders, and see the results immediately. By providing a graphical visualization of the database and search results, users can find trends and exceptions easily. User testing was done with eighteen undergraduate students who performed significantly faster using a dynamic queries interface compared to both a natural language system and paper printouts. The interfaces were used to explore a real-estate database and find homes meeting specific search criteria.
Here is a great talk by Ben where he states the goals of the visual information seeking process, and shows the Dynamic HomeFinder "thousand points of light" demo:
User Interface Strategies 1990 (University of Maryland Television Broadcast) Part 1-5:
In 1992 I was inspired to implement a version of it for SimCity I called the "SimCity Frob-O-Matic Dynamic Zone Filter", and I wrote about it a few years ago on Hacker News, and linked to a video demo:
DonHopkins on March 27, 2016 | parent | context | favorite | on: Direct Manipulation: A Step Beyond Programming Lan...
Ben Shneiderman has made a career of performing controlled experiments, measuring the efficacy of different techniques, comparing them to each other in different contexts, and teaching his students the importance of empirical testing, balanced with human centered design. [1]
He and Ike Nassi developed a goto-less visual programming technique called the Nassi-Shneiderman diagram [2]. He not only studied and summarized the status quo of flowcharting, but also conducted experiments that suggested flowcharts were not helpful for writing, understanding, or modifying computer programs.
He's also done a lot of work with information visualization [3], including tree maps and dynamic query sliders, and developed systems and published papers that have inspired many other people.
He and Christopher Williamson developed and empirically evaluated dynamic query sliders in the ingenious Dynamic Home Finder [4], which applies direct manipulation and infovis techniques to dynamic real time visual real-estate database queries. That inspired me to implement a similar real time information visualization technique in SimCity [5].
By studying, measuring, comparing, and generalizing on what was really going on, he came up with these eight golden rules for interface design:
1) Strive for consistency. Consistent sequences of actions should be required in similar situations...
2) Enable frequent users to use shortcuts. As the frequency of use increases, so do the user's desires to reduce the number of interactions...
3) Offer informative feedback. For every operator action, there should be some system feedback...
4) Design dialog to yield closure. Sequences of actions should be organized into groups with a beginning, middle, and end...
5) Offer simple error handling. As much as possible, design the system so the user cannot make a serious error...
6) Permit easy reversal of actions. This feature relieves anxiety, since the user knows that errors can be undone...
7) Support internal locus of control. Experienced operators strongly desire the sense that they are in charge of the system and that the system responds to their actions. Design the system to make users the initiators of actions rather than the responders.
8) Reduce short-term memory load. The limitation of human information processing in short-term memory requires that displays be kept simple, multiple page displays be consolidated, window-motion frequency be reduced, and sufficient training time be allotted for codes, mnemonics, and sequences of actions.
In my experience, age of homes in the neighborhood approximately equals age of kids in the neighborhood, +- a few years. And that makes sense if you think of who is typically buying houses en masse as a group - new families.
Now, I don't live somewhere like SF of NY where people cycle in and out, more cities of various sizes in the midwest and south.
My first house was built in the 40's. A few original owners existed, along with second owners. I noticed I guess people don't tend to move often. There were hardly any kids in the neighborhood.
Next house was built in 2009(this would have been 2018ish), and the neighborhood was packed full of kids.
Next house was built in the 80s. A lot of original owners, again, few kids.
Next house was built in 2012, this would have been 2021 - tons of kids
Next house built in mid 90s. This was 2022. Almost everyone in the neighborhood was the original owner, very few kids.
So, if my theory holds, if you want to find a neighborhood with a lot of kids, buy a midrange house in a ~5 year old neighborhood.
> I genuinely wish there was a map where you could see how many kids and their age groups are in a given neighborhood.
Maybe your local city/state/county/country government has a "Open Data" portal where they publish stuff like that? Barcelona for example has a pretty extensive Open Data collection (https://opendata-ajuntament.barcelona.cat/en) where you'd be able to find data like that (probably not ready made graphs/maps though), and probably also averaged data about how many people live in the households of a neighborhood, so you could extrapolate for families, etc.
If you are based in the UK, there is https://xploria.co.uk. Click/tap on any location on the map and you get a lot of information about the place, inclusive of generations, what percentage of families, singles, etc, schools within travel timw, and so on. Disclosure: I built this web app.
This is very cool - I haven't seen anything like it before. I've spent about half an hour noseying at places I've lived or considered living, fascinating. The view of changing house prices is something that I haven't seen presented in this way either.
Thanks for the kind words! The prototype was launched in 2013, you'd think the world would have caught up since :D. The platform underneath allows far more advanced functionality, so stick around, there will be many more usable (and useful) features in the near term.
For younger age groups, you could search for how many elementary schools are in the region. Our town has a lot of these schools, because people don't want to be sending their young children far away. Generally it's not until high school age where the students might have to travel farther from home to school.
That's exactly what I'm also looking for. I've been looking for a house in both Canada and the U.S. and for Canada there are some house search websites that show demographic census data for each neighborhood. For the U.S. I haven't seen anything comparable. I don't even know if this type of demographic data at the necessary granularity would be available publicly?
There is another website that I've been using that provides similar functionality in terms of the heatmap: http://close.city
What I like about this one is that it can show travel times from a specific address. What would be even more useful is if it could show mixed-mode transportation times.
Really puts into perspective how wildly varied cities in the US are in terms of density and development patterns. Compare New York [1] and Seattle [2]. Barely a single patch of gray in New York until you get to Long Island, while Seattle has big patches with no transit accessibility at all. And many of them are extremely wealthy areas. In New York, wealthier areas tend to have better transit. Very, very interesting.
This is cool but I don't think it fits my use case..
Seems like I have to pick criteria that have exact venues.. I want to pick abstract things like "walking distance from grocery", "biking distance from climbing gym" "1 hour drive to national park"
Hello!
To echo what loxias said, it is possible to make queries like this on the heatmap. You can use the "Search Nearby places" button - this takes in more general queries (like cafe, gym, walmart, etc) and gives you back a bunch of venues that fit that search.
This doesn't seem to work very well. Searching for "grocery store" netted me a set of 22 locations spread across the entire country.
Ah, you probably tried to search before the website got your location (so it couldn't bias your search).
I'll make this an impossible state, thanks for bringing this up.
I would like to not give my location to the website. Let me just pick a city manually.
That's fair!
See https://news.ycombinator.com/item?id=42978067
I don't really understand what the search nearby places button is doing. I think the solution would be to allow some sort of OR operator (which seems straightforward), or conditions the act on generic queries (which seems more difficult). For example say you live in NYC and want some nice green space near you. Choosing a single park is too narrow and choosing all parks is too broad. So you should be able to say e.g. "Central Park OR Prospect Park OR Brooklyn Bridge Park OR Fort Greene Park", or you should be able to say "Park, > 10 acres, 4+ star google rating, has tennis court, has bike path"... the point is that not all parks, grocery stores, coffee shops, etc are equal; I need to be able to qualify them somehow.
I missed this comment!
> Central Park OR Prospect Park OR Brooklyn Bridge Park OR Fort Greene Park
You can do this, actually. I kinda explain that here: https://news.ycombinator.com/item?id=42976053
The heatmap supports both AND and OR clauses
The second one (acrage, stars, etc) is harder, you're right.
The "Search nearby places" is really meant to be a convenience feature to fill your OR clauses. It works better for certain types of things. Like, sure, all parks is too broad because not all parks are equal. But use it for something like all Targets (the shopping chain) or something, and its more useful, since those are, more or less, all equal.
I'm still thinking of other convenience features for places that have more nuance, like parks.
You might prefer a more general walkability map, eg. https://www.walkscore.com
I found it was surprisingly easy to "populate with every instance of a given type". I've made a few maps based on grocery and transit where I did it by adding all >100 stores & stops.
"Total count of non-national-chain restaurants within a 1km walk" pretty much determines my criterion. Everything else is a once a week thing or less. The homes themselves are just a bunch of rooms, and I'm flexible. Quality of life is defined by my stomach.
Same, and I add one even more amorphous criteria, 'not directly next to noise pollution like an airport or highways.'
The available housing preferences may need to be reflected in the title.
My desired heatmap is for 5+ beds/3+ baths at [price range]. It's okay this isn't that - but the Housing Preferences descriptor indicates it might be.
I had the same impression. Like Price per Acre.
I didnt expect there to only be 1 type of constraint (travel time to a given location).
I think emphasizing its purely based on distance would be clearer.
Me too. I poked at the UX for a good few minutes trying to find out how to change the TYPE of constraint lol
I currently commute by train _and_ bike, but this only lets me filter by one or another, unfortunately.
This seems to be a common problem with navigation systems in general. It's easy to get walking+transit directions, but nigh impossible to get bike+transit, even though all the buses and trains near me let me bring a bike onboard.
Citymapper will do it. Haven't personally tried the results
Yeah the "mixed" navigation option sometimes provides it, but with many caveats:
- it never suggests bike+bus, only bike+train
- it will only ever try to put the biking on one end of the train ride, never both. I guess it assumes I'd be parking the bike at the station, not bringing it on board
- you can't actually "start" the navigation for some reason. It will just show the route overview
I have seen it schedule the bike at both ends. But it does seem like it assumes the bike is an 'option' like all the other forms of transport, which makes sense for bikes hired by the minute, but not if you're taking your own with you - you need all segments to be bike-friendly.
I put my wife's work location and my own in it and it correctly showed me that where we actually live would be a good location. When we moved to our house she worked at a different place, but I can see how this would be useful.
The "only show best matches" criterion is a little bit too aggressive in this case, though - it basically says "have you tried living in the middle of the highway"?
Also matched my place perfectly. With my work, leisure and commute requirements! Funny, I feel my requirements overfit on my habits - the area where I live was the only highlighted area with no alternative sweet spots.
For the nerds, the exact name for these heatmaps is Isochrone maps [0].
0: https://en.wikipedia.org/wiki/Isochrone_map
This is incredible! I've been looking at new houses for a while now, trying to balance living next to a park, walking distance to elementary school and a grocery store, and within a driving distance to my wife's work. I'm having a blast playing around with it right now and scoping out the potential neighborhoods that would be best.
One feature request: when you do "Search Nearby" (e.g. I did "elementary school") it found a _lot_ of schools, like 50 or more, well out side the city I was interested in. But short of adding them all and going through one by one and deleting them, is there a way to add just the matches in the city I put in the top-right corner? Maybe adjust the search radius or something? Or in the search results preview it gives you, a button to just select the few you want to add.
edit: oh, and another feature request. I'd love to accordion/collapse my criteria. Scrolling past my 40 parks to edit the bottom criterion is tedious.
I love the idea, but the data seems a bit off. I tested it in my little city, where I'm < 14 minutes by car from the grocery. It's actually a little quicker, but I'm going by Google Maps estimates.
Setting a criteria of 15 mins by car, I'm far out in the gray. I'd have to drive a couple miles to even get in the red. It's only 6 miles away!
> It's only 6 miles away!
Six miles is like a 30 minute drive when I am
Been there, done that, hopefully not again :).
I was hoping 'little city' would have indicated, but I should have specified, there is never enough traffic here to move the estimates much. Speed limit is 35ish the whole way.
Yeah, different geo analysis providers have different weightings for travel time (turns, street density, etc). It can sometimes lead to inconsistencies like this, especially if they don't use GPS and proprietary data to correct things...
TomTom maps (and probably others, I just worked for TomTom so know how theirs work) combine both real time traffic information, and road speed estimates based on data from users uploading their driving traces (with start and end locations removed to anonymize them), so it has a table of the average driving speed for all roads, on a per-hour and weekday/weekend basis, plus any real time traffic information it has, which it uses for route calculation and time estimation.
I want to live where I can walk to a grocery store, there is little traffic, crime and detritus, balancing affordability.
This doesn't cover any of that.
Why even comment then?
OP explained up front what the tool does
Its like walking by a poster advertising guitar lessons and exclaiming out loud "I don't want guitar lessons!"
Because I was induced to click by the title of the post:
"Show HN: A website that heatmaps your city based on your housing preferences"
I'm providing feedback that what is promised is not delivered. Maybe it's the op's preferences, but it's not my preferences.
It's more like a poster offering guitar lessons, and the person will only tell you that a guitar has five strings.
I thought about using Google maps street view photos for something like that, get all the photos for a city, and then classify how much trash is visible, and also calculate how many people there are and what they're doing.
My idea, at least from living in New York, is that if there are a medium amount of people walking around, and not that many people just sitting there or leaning against a building, and there isn't a ton of trash visible, then it is probably a pretty good neighborhood.
If there is nobody, or a ton of people, or if the people aren't mobile, or if there's a lot of trash, then it is probably not a neighborhood that I would like.
You could probably also use business types for that. Like, I would not want to live in a neighborhood that has a ton of pawn shops or churches in it.
Unfortunately if you built this, half of the world would publicly condemn you as a terrible horrible person.
I wanted to do this with a city I don't currently live in, but every time I did searches on specific locations it went for the nearest (bad) match to my current geographic location. It'd be nice to generalize it.
I do like that you used OSM rather than Google maps.
This is possible!
I should probably make the UX better. When you're on the heatmap page (or the distance matrix page), look at the top right of the screen. It shows you where it's basing its searches on, and you can override that bias with any location you like.
You only need to update it once per session - all pages and components will be updated
On the UX it would be helpful to add an option to delete all places.
It added 50 grocery stores with a "nearby places" search and inhad to refresh the page to start a new search instead of spending time clicking those X's.
His place finder is Google Maps. I know the guy that wrote it.
This is really cool! Well done. Like others, I'd love to specify generic things like distance to a "grocery store" or "gym" as part of my initial criteria. I see that I can add a long list of possible places that meet a search, so maybe I just want the UI to hide the details from me and add all those possible places for me.
I personally found the additional criteria being added to the top to be counter intuitive and I inadvertently deleted locations thinking it was the newest criterion, but it was actually my earlier ones. I think I've been trained to look/scroll to the bottom for the added element (e.g. like when adding additional Google Maps locations)
I would also love an option to mix transportation modes. For example, public transportation and biking.
Anyway, thank you for building this!
Very cool! This is nice when considering moving to a new area to narrow down neighborhoods that could work. One thing that I think could be useful is to also add criteria for things I want to be far from. Some people don't like living near airports for example.
Tom Carden made the "mySociety Travel Time and House Price Maps" in 2007:
https://www.tom-carden.co.uk/2008/01/24/mysociety-maps
>O'Reilly Radar has the scoop on the most recent thing I've finished working on at Stamen. Interactive travel time and house price maps for London. Go play, and read what mySociety have to say, including the ones for BBC TV Centre and the Olympic Stadium site. Then come back and read this full post if you want the background info...
O'Reilly Radar Article:
https://web.archive.org/web/20080208084133/http://radar.orei...
mySocieties: More travel-time maps and their uses
https://www.mysociety.org/2007/03/05/more-travel-time-maps-a...
This project became Mapumental (Mapumental was a mySociety project to plot journeys by time, not distance.):
https://www.mapumental.com
Unfortunately the live site was closed down, but the pages describe a lot of great inspirational ideas!
I'm not in the market for a new home right now but I would absolutely use this in a future search. As it is now, it helped me confirm that my house is in a great location for everything I do regularly.
It did seem to think that the closest "Bar" to me was a 19 minute drive, when in reality there are several within a 2 minute walk, however.
I've always liked the idea of combining distance criteria, especially when something is good for public transit to work, but not good in terms of walkability to anything I care about. (Some people walk for fun, I always seem to need an errand.)
I recall that Walkscore used to have something like this, and then it went away, and then it showed up on some other housing site... I was always surprised the type of feature didn't get more popular.
In terms of new features, there is a tricky problem of how to define things like "near a grocery store, the large kind, not that one tiny mini-mart". This brings in several overlapping challenges: How to get business locations and categorize them, how to allow the user to tweak that categorization or result, and how to efficiently turn a union of those the set of valid destinations into a combined region.
The "grocery store" problem is something I've been thinking about for a while, since it has two problems:
- There are tons of grocery stores (efficiency of processing all of those)
- Not all grocery stores are the same (supermarkets vs pricey luxury stores vs bodegas)
I've been thinking of mass processing one-time then allowing the user to super-impose pre-made heatmaps onto thier existing heatmap.
That would be exactly what I need!
Ideally something like 5min bike ride to the grocery store, 15min walk from a train station, and 30min drive to my in-law’s house.
It would be really interesting to do something like “10 minute bike ride to 3 or more grocery stores”. That would help reduce instances of niche specific stores, but also provides a much more useful variety.
I’d love to be able to find places that have 2+ or 3+ grocery stores within somewhat reasonable distance, and same thing goes for restaurants. Really any restaurant.
I'm imagining:
1. User defines a "multi-location" spec, like "MyFastFood" as "Having [2] or more of [Fast Food] excluding [Taco Bell,]"
2. User defines a requirement for their heat map which references the multi-location, ex: "Within [30 minutes] to [walk] to [MyFastFood]"
3. Within the context of a particular [user's requirement] and broad [city/town/region], a 2D area/gradient can be generated, and cached for a rather considerable period given how slowly businesses open/close.
Granted, that's the ambitious version. A simpler one would be to not support "at least X", and to combine the multi-location and the distance-rules all together into a single condition.
As others have mentioned, this kinda misses a "category" filter that is not just a search. I would suggest using the osm database for this (e.g. via overpass). If the data is good (depends on the location), this is better than just a keyword search.
I also have another issue: If I put the same thing in two different criteria (with different settings), it says the heatmap parameters are invalid.
My use case is this: I want to have a big station reachable within 30 minutes by public transit and any light rail station within 10 walking minutes. But this big station could fall into both of them, which, to me should then be handled according to the set criteria (maybe just treat it as two different entities entirely, just with the same coordinates?)
Noted, thanks!
> If I put the same thing in two different criteria (with different settings), it says the heatmap parameters are invalid.
That's odd, since the website does not differentiate places by coordinates. I think you might have been missing something else (like you clicked a "new place" button and didn't fill out that place, maybe)
Works for Sydney, well done!
I previously used Mapnificent to choose a house location, and found that many geographically closer properties were often much further in terms of time. Very useful. I like that it starts off with the map view, maybe do that in terms of “shortest time to a-ha” and to get the user into putting in the work to refine the output?
https://www.mapnificent.net/sydney/
I also used this to look at property/area info. Maybe lift some ideas off that. It used to have school ratings, crime etc.
https://heatmaps.com.au/
Something that’s a combo of both would be amazing. Good luck!
This is an excellent concept - and it may be hard to realize but it would be worth integrating "cost of living" and "languges spoken" concerns, and then to be able to find one's perfect spot on the planet.
I've been looking for a tool like this forever. Trying it for my city (Milan, Italy), and it's pretty accurate. Would it ever be possible to specify which type of public transportation to consider (ie: only metro, only bus, ...)?
Some suggestions:
- option to select individual results of "Search nearby places" or select top N closest result to your location
- collapsible criteria on the edit/view pages
- automatically name name untitled heatmaps. I have a ton of "Untitled Heatmaps" in my history which I forgot to name
- combine the edit/view pages - just have an edit page with a big "Generate" button so you can still edit criteria after heatmap generation
An easy way to make this great tool twice as good: Allow inverted selections.
I don't want to live near sketchy storefronts, and neither do other people less honest than myself.
The two-ended Min/Max sliders of the Dynamic HomeFinder and the SimCity Frob-O-Matic Dynamic Zone Filter allow you to effectively invert selections by dragging the Max end all the way to the right, and then adjusting the Min end, since the filter passes everything between Min and Max. Or you can drag the whole slider range by the middle to adjust Min and Max at once, exploring a fixed width interval of the values.
https://news.ycombinator.com/item?id=42978733
https://youtu.be/_fVl4dGwUrA?t=3m35s
A useful feature (not implemented in that video of X11 SimCity from 1992, but it would be easy to implement now with better graphics and faster computers) is to display a histogram in each of the sliders, where the x-axis is the parameter value, and the y-axis is the number of items of that slider's value (given that all other sliders are at their current value), so you can easily spot clusters and peaks and sparse areas, and you can include or exclude them from the filter by sliding the Min and Max edges across the histogram. So as you adjust one slider, the histograms in the other sliders change to reflect the current "slice" of multi dimensional space with respect to the filter you're adjusting.
Showing a histogram for every filter parameter on each slider gives you a multi-dimensional view of the distribution density of the data, that you can tweak and explore in real time, which helps you figure out how to adjust the filters to find interesting and ignore uninteresting items, and focus in on just the items you want.
I love histogram sliders! I think they were originally developed by Chris Ahlberg in Ben Shneiderman's group, and were later commercialized in Spotfire (I have yet to see them in any other infoviz system except Panopticon - https://www.perceptualedge.com/blog/?p=965). Jeff Heer did some work a few decades to generalize into 'scented widget' (http://vis.stanford.edu/papers/scented-widgets). With systems like DuckDB it's now even easier to implement them into various visual analytic systems.
I'm not sure why others have problems, but I love it!
Once, in a similar situation, I had the thought that "somebody should build that" and never imagined that somebody actually would.
With your tool, I learned that our home is better located than I thought and that there are many places more downtown that would be worse for me and my wife (bc of highways and good public transport connections).
I will show this some friends! Thanks!
You could take the aggregate of "where people want to live" and then compare that with an estimate of how much floor space the city allows to actually exist. I suspect the ratio would correlate extremely well with prices & rents per sqft of floorspace. In a perfect world this would inform urban planning, though in this world I am less confident.
This is really neat. It would be cool if you exposed some parameters like walking speed. This shows that a trip I make daily in ~12 minutes is just barely within the 30 minute walking shell from my house. Also if you want to waste an insane amount of time and effort, you could integrate this with a digital elevation model to yield better biking times.
Very nice potential in this idea. A few related use cases I would like:
- population density
- average cost per square foot
- distance from river or lake (I live in Chicago)
This is cool!
It seems nobody has asked/suggested this so I'll do it - I would love to specify time of day/traffic conditions during which to determine time taken to travel.
In the next year or two we'll be looking to move and since we work on the southern side of the city, living out to the north wouldn't work due to traffic through the city.
This is so useful! I have wanted a tool like this for years. Thank you for putting in the work and sharing.
I like the heat map. Years ago I made something similar with only generic categories like bakery or shop. In the end it wasn't very useful though.
https://homeground.qwoot.net/
This is really cool (and timely for me). Lovely work with the UX. No accounts, no nonsense. Kudos.
I've built a similar app, but for the whole US, where you can input your personal preferences and find the areas in the US that best match based on the data:
https://exoroad.com
I tried adding my city's main bike path that goes around the whole city and it seems to only register it as one point instead of a bunch of access points unless I go look up every entrance and manually add it.
Would be nice to use conditions. Like, if there are multiple libraries in a city and I would like to evaluate being 15 minutes from any of them, have a condition able to be set -
15 minutes from: Library A OR Library B OR Library C
That actually is possible
Heatmaps are split into criteria, and each of those criteria can contain multiple places.
The criteria are OR clauses between any of the places, and the heatmap is an AND clause over all the criteria
So to do what you'd like, you'd place the libraries in the same criteria
This would be great if I had enough money for my preferences to matter
Everyone's preferences matter.
I know a homeless guy in Texas whom I talk to on IRC daily. Even he has preferences. He lives under a bridge near a library that gives him Internet access.
Is the only criteria travel times? I couldn't find others. Usually there's a mix of criteria that are vital when house hunting?
This is fun, living in NYC I need the opposite as well –
1. At least 2 blocks from a police station
2. At least 2 blocks from a fire station
3. At least 1 block from a bar
there was this last year, about isochrone maps:
https://news.ycombinator.com/item?id=39808215
not sure how these apply to variants like 10mins by feet vs by 10mins bike vs by bus vs by metro vs by car vs by airplane :)
This is fun - waiting so long for the map to generate is a bit of a drag, but understandable.
I have a v2 of that algorithm coming this weekend! I expect it to make it MUCH faster :)
I believe these are called contour maps not heatmaps (I might be wrong)
You should see if you could make this an extension to overlay on Zillow
It can! I can overlay with Zillow, Apartments.com and Streeteasy
https://theretowhere.com/we-got-an-extension-babyyyy
Great idea! But unfortunately the site was too hard for me to use on mobile (iOS, Brave browser). I was unable to set the location - when I tried to set the location, the screen would randomly zoom in and I couldn’t scroll through the addresses.
It only cares about distance by car, walk, etc?
I would like to have more criterias
This doesn't work for Tokyo or Taipei currently :(
Great idea, looking forward to someone implementing it!
This is really, really cool and thanks for making it!!
We must think at least somewhat similarly, last few times I was apartment hunting I did the same, though I never polished it up like this (more plugging numbers into a spreadsheet).
Honestly the biggest thing this does for me is validate that the data APIs must exist for what I'd really want, which is write something to make much larger and more complex "programmatic" maps -- the list of places being generated by a more complex sequence of steps for instance, and the combining function for different criteria including nonlinearities.
Curious how you're computing the walking distances, I'm guessing this is combining some off the shelf API for it with another for the points of interest? Though it would be badass if you did it from scratch starting from just OSM. ;)
What if I’m a Kaiju, and I want to disrupt the most electrical cables juice up? What path of destruction is best for me?
See also, Hood Maps [0], a crowdsourced map of annotations of people's opinions on particular areas in a city. It's by Pieter Levels who built Nomad List, PhotoAI, etc, you might know him if you're in the indie hacker and startup space at all.
[0] https://hoodmaps.com
I tried something similar myself but failed on wanting to load the whole world of openstreetmap onto my server and into postgres. It worked but I never managed to get sub second postgres query times on a billion rows. Hence, making it unsuable for any user.
So my question to you. What's your server and data setup? Do you even have your own data? I'm very curious on what is actually needed to make it work anywhere.
I’ve been looking for something like this. On my phone it’s rather unclear how to interpret the heat map though. There’s some kind of broken tooltip. What do the colors represent? It looks like it would be the distance to the closest point rather than to all the points?
How do you compute the heat map? Thousands of API requests from various points?
It reminds me of the "Dynamic HomeFinder", developed by Christopher Williamson and Ben Shneiderman at the University of Maryland Human Computer Interaction Lab in the late 80's, which they published a paper about in SIGIR '92: "The Dynamic HomeFinder: Evaluating Dynamic Queries in a Real-Estate Information Exploration System".
https://web.archive.org/web/20161119140228/http://hcil2.cs.u...
>Abstract: We designed, implemented, and evaluated a new concept for visualizing and searching databases utilizing direct manipulation called dynamic queries. Dynamic queries allow users to formulate queries by adjusting graphical widgets, such as sliders, and see the results immediately. By providing a graphical visualization of the database and search results, users can find trends and exceptions easily. User testing was done with eighteen undergraduate students who performed significantly faster using a dynamic queries interface compared to both a natural language system and paper printouts. The interfaces were used to explore a real-estate database and find homes meeting specific search criteria.
Here is a great talk by Ben where he states the goals of the visual information seeking process, and shows the Dynamic HomeFinder "thousand points of light" demo:
User Interface Strategies 1990 (University of Maryland Television Broadcast) Part 1-5:
https://www.youtube.com/watch?v=Yz_7nJN8jk0
In 1992 I was inspired to implement a version of it for SimCity I called the "SimCity Frob-O-Matic Dynamic Zone Filter", and I wrote about it a few years ago on Hacker News, and linked to a video demo:
https://news.ycombinator.com/item?id=11370099
DonHopkins on March 27, 2016 | parent | context | favorite | on: Direct Manipulation: A Step Beyond Programming Lan...
Ben Shneiderman has made a career of performing controlled experiments, measuring the efficacy of different techniques, comparing them to each other in different contexts, and teaching his students the importance of empirical testing, balanced with human centered design. [1]
He and Ike Nassi developed a goto-less visual programming technique called the Nassi-Shneiderman diagram [2]. He not only studied and summarized the status quo of flowcharting, but also conducted experiments that suggested flowcharts were not helpful for writing, understanding, or modifying computer programs.
He's also done a lot of work with information visualization [3], including tree maps and dynamic query sliders, and developed systems and published papers that have inspired many other people.
He and Christopher Williamson developed and empirically evaluated dynamic query sliders in the ingenious Dynamic Home Finder [4], which applies direct manipulation and infovis techniques to dynamic real time visual real-estate database queries. That inspired me to implement a similar real time information visualization technique in SimCity [5].
By studying, measuring, comparing, and generalizing on what was really going on, he came up with these eight golden rules for interface design:
1) Strive for consistency. Consistent sequences of actions should be required in similar situations...
2) Enable frequent users to use shortcuts. As the frequency of use increases, so do the user's desires to reduce the number of interactions...
3) Offer informative feedback. For every operator action, there should be some system feedback...
4) Design dialog to yield closure. Sequences of actions should be organized into groups with a beginning, middle, and end...
5) Offer simple error handling. As much as possible, design the system so the user cannot make a serious error...
6) Permit easy reversal of actions. This feature relieves anxiety, since the user knows that errors can be undone...
7) Support internal locus of control. Experienced operators strongly desire the sense that they are in charge of the system and that the system responds to their actions. Design the system to make users the initiators of actions rather than the responders.
8) Reduce short-term memory load. The limitation of human information processing in short-term memory requires that displays be kept simple, multiple page displays be consolidated, window-motion frequency be reduced, and sufficient training time be allotted for codes, mnemonics, and sequences of actions.
----
[1] An Empirical Comparison of Pie vs. Linear Menus: http://www.donhopkins.com/drupal/node/100
[2] Nassi-Shneiderman diagram: https://en.wikipedia.org/wiki/Nassi%E2%80%93Shneiderman_diag...
[3] InfoVis: https://en.wikipedia.org/wiki/Information_visualization
[4] Dynamic Home Finder: http://hcil2.cs.umd.edu/trs/92-01/92-01.html
[5] SimCity Frob-O-Matic Dynamic Zone Filter: https://youtu.be/_fVl4dGwUrA?t=3m35s
I genuinely wish there was a map where you could see how many kids and their age groups are in a given neighborhood.
Our current house is great, but there aren't many kids in the neighborhood.
I understand this is sensitive information, so it probably doesn't exist.
But choosing a neighborhood with other families that are in a similar life experience is kind of hard..
Especially considering it seems that kids play outside much less, so that's less of a signal.
In my experience, age of homes in the neighborhood approximately equals age of kids in the neighborhood, +- a few years. And that makes sense if you think of who is typically buying houses en masse as a group - new families. Now, I don't live somewhere like SF of NY where people cycle in and out, more cities of various sizes in the midwest and south.
My first house was built in the 40's. A few original owners existed, along with second owners. I noticed I guess people don't tend to move often. There were hardly any kids in the neighborhood.
Next house was built in 2009(this would have been 2018ish), and the neighborhood was packed full of kids.
Next house was built in the 80s. A lot of original owners, again, few kids.
Next house was built in 2012, this would have been 2021 - tons of kids
Next house built in mid 90s. This was 2022. Almost everyone in the neighborhood was the original owner, very few kids.
So, if my theory holds, if you want to find a neighborhood with a lot of kids, buy a midrange house in a ~5 year old neighborhood.
(and yes, I realize I've moved a lot).
> I genuinely wish there was a map where you could see how many kids and their age groups are in a given neighborhood.
Maybe your local city/state/county/country government has a "Open Data" portal where they publish stuff like that? Barcelona for example has a pretty extensive Open Data collection (https://opendata-ajuntament.barcelona.cat/en) where you'd be able to find data like that (probably not ready made graphs/maps though), and probably also averaged data about how many people live in the households of a neighborhood, so you could extrapolate for families, etc.
If you are based in the UK, there is https://xploria.co.uk. Click/tap on any location on the map and you get a lot of information about the place, inclusive of generations, what percentage of families, singles, etc, schools within travel timw, and so on. Disclosure: I built this web app.
This is very cool - I haven't seen anything like it before. I've spent about half an hour noseying at places I've lived or considered living, fascinating. The view of changing house prices is something that I haven't seen presented in this way either.
Thanks for building it and thanks for sharing.
Thanks for the kind words! The prototype was launched in 2013, you'd think the world would have caught up since :D. The platform underneath allows far more advanced functionality, so stick around, there will be many more usable (and useful) features in the near term.
Is this based on census data?
Yes, although not yet updated to the one of 2021 (2022 in Scotland).
The main page has references to all data sources.
For younger age groups, you could search for how many elementary schools are in the region. Our town has a lot of these schools, because people don't want to be sending their young children far away. Generally it's not until high school age where the students might have to travel farther from home to school.
That's exactly what I'm also looking for. I've been looking for a house in both Canada and the U.S. and for Canada there are some house search websites that show demographic census data for each neighborhood. For the U.S. I haven't seen anything comparable. I don't even know if this type of demographic data at the necessary granularity would be available publicly?
There is another website that I've been using that provides similar functionality in terms of the heatmap: http://close.city
What I like about this one is that it can show travel times from a specific address. What would be even more useful is if it could show mixed-mode transportation times.
Really puts into perspective how wildly varied cities in the US are in terms of density and development patterns. Compare New York [1] and Seattle [2]. Barely a single patch of gray in New York until you get to Long Island, while Seattle has big patches with no transit accessibility at all. And many of them are extremely wealthy areas. In New York, wealthier areas tend to have better transit. Very, very interesting.
[1]: https://close.city/?x=-73.92368&y=40.74092&z=11.62779&r=0&l=...
[2]: https://close.city/?x=-122.32293&y=47.63375&z=12.62654&r=0&l...
Goes to show the benefit of a well connected subway network. Seattle is mostly just buses except for two independent light rail lines