Tumblelog by Soup.io
Newer posts are loading.
You are at the newest post.
Click here to check if anything new just came in.

May 26 2018

OpenStreetMap User's Diaries: Mapping Tanzania and Nigeria

Working on #4358 - Tanzania Development Trust: Mwanza West Community Mapping Project as a task for LetGirlsMap. I have also made contributions to task #3584 - Mapping for Flooding Emergency Intervention and Support #NRCSkano in Nigeria.

OpenStreetMap User's Diaries: Will the DWG block us all one day? - Part II

As recently suggested in my previous post "Will the DWG block us all one day?" I took a deeper look into it. I have scraped all 1934 user blocks available last friday together with some user information.

As already indicated, the DWG is blocking more and more user (< no judgement) .... probably along with the increasing interest in OSM and the influx of trolls and paid editors.


If the trend continues, 2018 will be another record year of user blocks.

MedianBlocks On the other hand, the median block duration is at a very low level. (in case someone does not like annualisation, here is a YTD May chart.

If you look who is blocked, you can see that the vast majority are newbies. A mapper is considered as new if the account age at block date is less than 6 months. (If one is blocked more often, he could fall in both categories) However, experienced mapper blocks are also on the rise slowly on average. BlocksByType

The average block duration seems to be unrelated to the mapper type. The spike in 2014 is mainly caused by the 100 years block of user Sorein. In 2016 emacsen blocked some new users for 10 years which resulted in the other spike. Example


Just in case there are some troubles with the DWG (I am looking into it for a friend), you may want to know whom to avoid.


DWG stand for the OSMF Data Working Group account They blocked 6 accounts for 9 years, all belonging to one criminal case. The other outlier is emacsen, which -as mentioned earlier- blocked someone forever. So let's leave this aside.


In case you are in trouble, you should avoid Peda, SomeoneElse and woodpeck if you are a new mapper. Try pnorman instead. If you are already a while around, SomeoneElse seems to be a better choice than pnorman and woodpeck.

OpenStreetMap User's Diaries: Drinking water fountains.

It is the time of the year to check them.

OpenStreetMap User's Diaries: Miami Beach Real Estate ! Douglas Elliman

Luxury Miami Penthouses for Sale Rent ! Miamibeachhomefinder.com

A South Beach penthouse located at 321 Ocean Drive once listed for listed for sale for $53 million just sold for $26 million. At one time this was South Beach’s priciest residence available for sale. The sellers re listed PH with Douglas Elliman and reduced the asking price to $34 million in August 2017. As per records, the sellers purchased the Miami Beach penthouse at Pre-construction for $20 million in 2013.

OpenStreetMap User's Diaries: Mapping with OpenStreetMap in Davao City

It's been almost a week since I got back from Davao, but I've yet to write anything down about the OpenStreetMap activities we had there. So, here's a quick round-up:

"Open Source Mapping" with Tambayan Center

"Is the Bermuda Triangle for real?"

This is the most interesting question I got from this workshop!

This mapping project was in the pipeline since the middle of 2017 when Tambayan Center sent me a query over email, and it just happened that I was also in Davao at that time. I dropped-by (and, of course, mapped ) their office to get their feet wet about OSM, and maps in general, and listen to details of what they're trying to do. Tambayan Center is a long-standing NGO operating in Davao and working mostly with concerns and rights of young people, and their communities.

image Breaking ~~bread~~ ice cream and chips with the Tambayan Center team during my last visit

It was fun working along-side Jen, and engage a younger set of volunteers (ages: 11-18) with a fresh style. She's also passionate mom, and advocates for womens' rights in various ways, other than mapping.

image Tambayan Center workshop activities: Editing, Field work, Community validation, Fun

Jen will be posting a more detailed write-up soon, and I'm looking forward to seeing that. I hope you do, too.

Mapping with OpenStreetMap x Ateneo de Davao University

Mr. Glenn Depra of Ateneo de Davao University organized a workshop for faculty members of the university, with their Social Research Training and Development Office hosting the affair. This was originally intended as a workshop that would be made open to the public, but due to last minute hitches, we decided to defer offering it for another time.

The workshop included a general coverage of available (Free software), editing tools, data collection , and how to make use of OSM data, including Q&A segments for specific concerns.


Some of the participants were also from a group people Maning Sambale met during a mapping party back in 2009. They are surprised to find out that many of major map features added on OSM stems back from that mapping party made almost 9 years ago.

The Ateneo community is enthusiastic about helping promote OpenStreetMap (and Open data) not just for their academic pursuits, but also to engage other communities and organizations in Davao.

Instant meet-up with a local mapper

image With my luggage in tow, coming in late from another meeting, and off to another afterwards.

Whenever I'm tramping about, I try to connect with local mappers, and chat about maps, open data, and free software. Of course, this often leads to (hours of) talking about so many other things. It's good meeting you, Alex, and I hope we'd see you in another workshop soon.

May 25 2018

Changeset in subversion [34218]: PointInfo: Adds a link to photo metadata in Catastro module. Patch by ...

PointInfo?: Adds a link to photo metadata in Catastro module. Patch by Javier Sánchez Portero.

OpenStreetMap User's Diaries: Alternative Routes

In the following I describe how we implemented reasonable alternative routes in the Open Source Routing Machine's Multi-Level Dijkstra mode.

Final result: generating reasonable alternative routes incredibly efficiently

Our main goals when designing and implementing reasonable alternative routes were two-fold:

  • provide reasonable alternative routes for end-users and use-cases such as generating many alternative routes letting users or developers decide between them (e.g. runner's app, start and end is the same but provide different route each day), and
  • guarantee for fast query times in the tens to lower hundreds of milliseconds to fit the Open Source Routing Machine's trade-off of doing heavy pre-processing and then get incredibly fast exact responses

But what exactly do we mean by "reasonable" alternative routes? When we find a fastest route and compute alternatives these are the criteria we try to optimize for:

  • the alternative route must not be x% longer than the fastest route; it also depends on the route's length (think: 10 minutes vs 13 minutes is fine, 10 hours vs 13 hours is not)
  • the alternative route must be x% different than the fastest route
  • the alternative route must be "locally optimal": x% of the detour is a fastest route itself (think: if the alternative route takes a different highway than the fastest route then you're on the fastest route to your destination on this highway)

We can tune these thresholds to either make the alternatives look better or return more alternatives to the user.

There are subtle but essential differences in what we implemented compared to what users might expect:

  • we do not specifically return a route e.g. using highways and a route avoiding highways. The avoid-x feature works indepedently of the alternatives implementation and users can make two requests: one avoiding highways and one using highways.
  • we do not return a route e.g. based on fastest time and a route based on shortest distance. Our alternative routes are fastest routes taking traffic and other penalties into account. Different metrics still need different graphs (profiles) until support for multi-metric lands.

Here's the general idea for the so called Via-Method we implemented for finding alternative routes for a s -> t request:

  • start bi-directional Dijkstra forward search from s. Save all vertices
  • start bi-directional Dijkstra reverse search from t. Save all vertices
  • when both search spaces meet (and you usually would stop and reconstruct the shortest path) continue both searches "for a bit"
  • intersect both vertex sets: these are the candidate vertices
  • for all candidate vertices c a (potentially arbitrarily bad) alternative route is (s, c, t)
  • evaluate and filter alternative routes based on stretch, overlap, how "reasonable" it is

Note: for different techniques check the extensive list of resources linked at the end.

Here is an example for the set of alternative routes you get initially without any pruning:

The main problem here is: the candidate set is huge and needs to be pruned for reasonable candidates.

In addition the filtering needs to be fast and happen gradually: to stay within the time bounds we can not e.g. extract and evaluate the detailed path of every single alternative candidate. We need to filter alternative candidates down step by step, with cheap high-level criteria first, and expensive detailed criteria last.

For the initial via-vertices we make use of the Inertial Flow graph partition (osrm-partition). The graph partitioner recursively (and fully in parallel) cuts the graph "in half" using Dinic's algorithm for the min-cuts.

Here is the search space we then take via-vertices from to construct alternative route candidates. We know all routes from s to t have to go through these candidate nodes due to the partition:

When we then use the partitioner's lowest level cells to filter out alternative route candidates that are "too similar" we already can see improvements:

We then focus on local optimality. The overall problem we try to solve is as follows: we could pick a via node for which the alternative path would make an unreasonable detour. Think of picking a via node which leads to a route leaving the highway and then immediately going onto it again.

To guarantee for local optimality around the via-vertices we take a look at the two sub-paths around a via node and use the forward and reverse heaps to check if they are optimal.

Local sub-path is optimal:

Only then we unpack the remaining alternative candidates for their detailed paths and run extensive high-quality evaluation checks resulting in these results:

Last but not least all of the above has to happen on all pair-wise alternative routes: when we return multiple alternative routes they not only have to be reasonable compared to the shortest path but to all other alternative routes themselves, too.

Give it a try, happy to hear your feedback!

I would like to thank Moritz Kobitzsch who literally has a PhD in alternative routes and was sitting across my desk guiding this adventure!

If you are interested I encourage you to take a look at the original ticket, pull request and implementation. Also check out the extensive list of resources below for further details.


OpenStreetMap User's Diaries: Alternativrouten

Im Folgenden beschreibe ich wie wir Alternativrouten in den Multi-Level Dijkstra Algorithmus der Open Source Routing Machine implementiert haben.

Endprodukt: effiziente und plausible Alternativrouten

Unser Fokus beim Designen und Implementieren von Alternativerouten lag dabei auf:

  • hoch-qualitative Alternativerouten zu generieren: fuer den Endanwender und fuer Anwendungsfaelle die von vielen Alternativerouten profitieren koennen (z.B. eine Jogging-App bei der Start und Ziel ueber die Woche hinweg gleich bleiben aber die Route sich aendern soll), und
  • gleichzeitig fuer schnelle Antworten der Routingengine garantieren zu koennen, die im Bereich von 10-100 ms liegen und damit in das Konzept der Open Source Routing Machine passen wobei pre-processing benoetigt wird aber dafuer extrem effizient und schnelle exakte Routen berechnet werden koennen.

Hier sind die Kriterien nach denen wir die Qualitaet der Alternativerouten beurteilen:

  • die Alternativroute soll nicht laenger als x% verglichen zur schnellsten Route sein; das haengt natuerlich auch von der Laenge der Route ab (z.b. ist eine Alternativroute von 13 Minuten okay wenn die schnellste Route 10 Minuten betraegt; 13 Stunden Alternativerouten bei 10 Stunden schnellste Route ist nicht mehr okay fuer den Nutzer)
  • die Alternativeroute soll sich um mindestens x% verglichen zur schnellsten Route unterscheiden
  • die Alternativeroute soll "lokal optimal" sein: x% der Alternativerouten-Umweg ist wiederum eine schnellste Route

Wir koennen an diesen Parametern nun so drehen dass wir entweder Alternativerouten hoeherer Qualitaet bekomme oder generell mehr Alternativerouten potentiell schlechterer Qualitaet.

Was wir fuer diese Implementierung nicht betrachten:

  • wir geben dem Nutzer keine explizite Mischung an Alternativerouten bestehend z.b. aus "Nutze Autobahn" und "Vermeide Autobahn". Dafuer gibt es ein unabhaengiges Feature in der Routing engine.
  • wir geben dem Nutzer keine multi-metric Alternativerouten, also keine Alternativerouten basierend z.b. auf Zeit, Distance, und Aehnlichem. Die Alternativerouten basieren auf einer einzigen Metric die von den Profilen und Traffic Daten bestimmt wird.

Hier ist der grobe Ablauf beschrieben wir wir Alternativerouten mitterls der so genannten "Via-Methode" berechnen:

  • starte die bi-direktionale Dijktra Vorwaertssuche von s; speichere alle besuchten Knoten
  • starte die bi-direktionale Dijktra Rueckwartssuche von t; speichere alle besuchten Knoten
  • wenn sich beide Suchraeume treffen (und die Suche normal abgebrochen werden kann, da eine schnellste Route gefunden wurde), lasse beide Suchen "ein bisschen" weiterlaufen
  • schneide die zwei Mengen an Knoten: das sind die Knoten fuer die beide Suchraeume sich ueberlappen.
  • fuer alle dieser Knoten c kann eine (potentiell schlechte) Alternativeroute erstellt werden ueber die Route (s, c, t)
  • evaluiere und filtere diese Alternativerouten Kandidaten basierend auf den oben genannten Kriterien

Im Folgenden schoen zu sehen wie diese Alternativerouten Kandidaten aussehen:

Das Hauptproblem ist nun dass es etliche Alternativerouten Kandidaten gibt die gefiltert werden muessen.

Das Filtern muss also extrem effizient und schnell sein. Aus diesem Grund filtern wir schrittweise, von inexakten groben Filtern am Anfang die z.b. nicht die genaue Geometrie entpacken muessen, zu hoch-qualitativen Filtern die wegen Laufzeitbeschraenkungen auf nur noch wenigen Alternativerouten Kandidaten laufen koennen.

Die initialen Via-Knoten bekommen wir mittels Intertial Flow Graph-Partitioniers (osrm-partition). Der Graph-Partitioniere schneided den Graph rekursiv (und komplett parallelisiert) mittels Dinic's Min-Cut Algorithmus in zwei Stuecke.

Hier ist der Suchraum fuer eine Route Muenchen -- Berlin aus dem wir die Via-Knoten fuer die Alternativerouten Kandidaten entnehmen. Dank des Graph-Partitioniers wissen wir dass alle Routen durch diese Via-Knoten gehen muessen.

Wir koennen dann die untersten Level der Partition verwenden um Alternativerouten Kandidaten zu filtern die "zu aehnlich" sind, ohne dass wir die Geometrien der Routen entpacken muessen (was extrem teuer waere).

Ein weiteres Kriterium fuer die Qualitaet der Alternativerouten ist die so genannte Lokale Optimalitaet. Das Problem das Lokale Optimalitaet versucht zu loesen sind Alternativerouten die kurzfristig einen unzumutbaren Umweg nehmen. Das koennte beispielsweise so aussehen, dass die Alternativeroute von der Autobahn kurz ab geht, nur um dann kurz danach wieder auf die Autobahn zu gehen.

Um fuer Lokale Optimalitaet zu garantieren schauen wir uns die Routen um die Via-Knoten an und nutzen die Heaps der Forwaerts- und Rueckwartssuche um zu testen ob diese Routen selbst schnellste Routen sind.

Die lokale Route ist selbst optimal:

Erst jetzt entpacken wir die Geometrien der Alternativerouten-Kandidaten und ueberpruefen unsere Kriterien auf den detaillierten Routen.

Und zuletzt muessen wir diese Kriterien nicht nur zwischen der schnellsten Route und den Alternativerouten-Kandidaten ueberprufen, sondern sobald wir mehrere Alternativerouten gefunden haben gegen alle paarweise Alternativrouten selbst.

Probier's aus mit der Open Source Routing Machine - wir sind offen fuer Feedback!

Ein besondere Dank geht an Moritz Kobitzsch der eine grosse Hilfe war und mit seinem Doktor in Alternativrouten mit seinem Wissen extrem helfen konnte!

Mehr Details:

OpenStreetMap User's Diaries: 计划01

  • 1.济南经十路-山师东路-和平路-历山路-花园路-花园东路-奥体西路-经十东路片区初步添加完善。
  • 2.济南43-50路公交车,127-135路公交车,BRT3-5路公交车路线,之前编辑过的济南公交车路线修改。
  • 1.1目前计划,山大路-解放路-闵子骞路-益寿路初步添加完善。
  • PS:如有不足请指出更正,最好不要把建筑物或区域的节点连在道路上。

OpenStreetMap User's Diaries: Best Garage Doors Cotswolds

Find the top quality and most durable roller, Garage doors Cotswolds at Cheltenham Roller Doors! We provide the wide range of roller and garage doors at a very reasonable price. We also provide the services of cover repairing, door repairing, maintains, Roller shutter doors supplying and fitting etc. In Cotswolds. Get in touch.

OpenStreetMap User's Diaries: Interior Designers Christchurch

Partners in Design is a top quality interior designers Christchurch that can help you to select the best & unique interior design and make them true for you. We provide you the best services according to your budget & expectations. Our expert designers help you to improve your home, room or any space. Contact us now!

OpenStreetMap User's Diaries: Docupile

Docupile is a quick and easy to use software. It stores all the documents securely and allows us to perform a quick search of our data whenever we need it. Overall its a great application for the price we pay!

OpenStreetMap User's Diaries: Active Care Solutions Limited

Children need the right support to flourish and at Active Care Solutions we believe that supporting a child’s culture and religion whilst they are in foster care is incredibly important in offering the best support. Call us on 0800 917 7937 for more information about how you can help support a child in foster care. Active Care Solutions, Pure Office, Broadwell Road, Oldbury https://www.activecaresolutions.co.uk/

OpenStreetMap User's Diaries: Foster Enterprises

Foster Enterprises is the leading custom scarf manufacturers and suppliers in UK that provide the wide range & design of high quality scarves at very reasonable prices. We provide you the exact value of your money and help you with the best design according to your thinking. Contact us now!

OpenStreetMap User's Diaries: Why it is important to have a Digital Marketing strategy in an online business

Evolution of Digital Marketing In 2012, digital marketing had taken off with many risks which have shown an improvement over the years. In this, a digital marketing training is a necessity which had improved the situation further. As best digital marketing courses started to teach people about the advantages of digital marketing and their effects on the society, the trend started to grow more and more. In fact, these trends are making a huge profitable impact on businesses in well-developed cities like Bangalore with best digital marketing courses in Bangalore being offered. However, initially, losses were incurred which set a stepping milestone for the digital marketing career. Now, it is a tradition to market products and ideas online.

Approach to Develop Strategy for Digital Marketing In order to pursue Digital Marketing, an approach should be planned. It is necessary to understand a strategy in order to carry forward the business for a longer period of time. However, considering this needs awareness. That is why a digital marketing course is necessary to go over the various modules of digital marketing. Hence, it is necessary to go over some points which may be required to plan the digital marketing strategy.

Aligning digital marketing with the agenda In order to reach business goals, digital marketing might be a necessary step. Focusing on acquiring customers or introducing products to the market, even brand awareness building requires the step by step guide in developing a digital strategy. In order to focus on the end results to perform efficient marketing, one needs to focus on target audience, the message, and the persuasion strategy.

Focusing on the important points

Digital Marketing Institutes spread awareness about the work related to the process, from data analysis to development of the content. The team might focus on various other aspects however, there are certain activities which require must attention. In order to set the workflows of a critical nature, a strategy needs to be followed.

Importance of technology Marketing platforms to tools for analysis, are all considered to be a part of digital marketing strategies. A clear strategy is required to perform these tools to invest in technology to expect returns which come more than the base expectations. This enables in delivering customer experiences.

Required resources Digital marketing requires many aspects including human resources, financial investment, and technological resources. Accomplishing the set goals becomes easy when a strategy is followed, it is like a target which gets accomplished on various levels. These strategies must be brought into the company's goals, whether the time of the staff is a requirement, IT or a huge budget. Digital Marketing courses in Bangalore provide optimum training on various factors of strategic planning, particularly with respect to the resources planning and management. Such strategies help the entrepreneurs to devise a methodical approach to the staff requirement or need to access critical tools.

Focusing on the customers Understanding the requirements of the customer helps in hitting the right pulse. However, huge data availability has restricted the selection of any clear insight. A strategy devised for digital marketing helps the companies to achieve key performance indicators, methods on collecting data and systems related to technology which will be built to offer customers with profound information.

Minimization of duplicates Creation of content on various aspects of the company might make digital marketing complex. Content creation, social media marketing and other ways to increase the flow of information requires being followed by a strategy. This will prevent wastage of resources and effort duplication. Following a set pattern will focus on the right requirements of the digital marketing aspects and this can be well achieved through best training. As an example, digital marketing courses in Bangalore offers an in-depth training focussing on strategies like content creation, social media and so on.

Valuable experience by customer Digital marketing helps companies in building a nice customer base and their experiences. A strategy will determine the experience of the customers and the roadmap to achieve it. Using an Omni-channel interface helps customers to experience digital marketing at the user end by filling the gaps in a disciplined method.

Experimenting and performance improvement Experimenting the right results help in understanding the progress over time. The usage of the right resources involves a strategic output. It is easy to place the resources correctly if the path is known. In order to experiment with the mind of the customer's, call to action helps in the proactive participation. A digital strategy will help in realizing the ins and outs required for building the brand.

Conclusion Learn Digital Marketing courses in Bangalore, because Digital marketing is now a trend on demand. Digital marketing courses holds a supremacy and proclaim a vision to reach the need of the audience. It has been made a compulsory method for attracting customers towards the brands, by offering them with the correct information and providing equal support and interaction. Brands which start off without a strategy fail to make an impact on the digital marketing front. Devising pre-planned approaches help in goal definition, resource focusing and yielding results from the objectives defined for the business. The efforts will pay off naturally.

Keywords are #Digital marketing courses in Bangalore, #Digital marketing courses in Bangalore with placements, #Digital marketing courses in Bangalore fees, #SEO courses in Bangalore.

OpenStreetMap User's Diaries: Canvasman

Canvasman is one of the leading companies that provide outdoor covers at very affordable prices. We help you with boat covers upholstery, shade sails, shop canopies etc at very reasonable prices. For more info visit our website.

May 24 2018

OpenStreetMap User's Diaries: The power of OSM

I would like to recount an interesting encounter I had during a bike trip in Myanmar. In December 2016 I wanted to cycle from Yangon/Myanmar to Kuala Lumpur/Malaysia. Beforehand I was armchair mapping some tracks east of Bago, Myanmar in order to avoid the stinky and noisy main road and slightly short-cut the trip. Later I used OsmAnd+ to navigate those tracks.

It was precisely on these roads in the middle of nowhere - Google Maps is literally blank there - where I met a female solo bike traveller. She used Maps.me to navigate and was using 'my' roads to go into the opposite direction.

Quite entertaining to see the real world impact of all our mapping efforts.

Older posts are this way If this message doesn't go away, click anywhere on the page to continue loading posts.
Could not load more posts
Maybe Soup is currently being updated? I'll try again automatically in a few seconds...
Just a second, loading more posts...
You've reached the end.

Don't be the product, buy the product!