[Talk-GB] OSM University of Liverpool exercise (or 200 free(ish) volunteers)

SK53 sk53.osm at gmail.com
Thu Nov 27 21:09:37 UTC 2014


All my metrics are based on survey followed by data entry. I am sure direct
entry using OsmAND or Vespucci has a significant advantage of avoiding the
some of the work involved in the data entry phase. Unfortunately, using the
GPS on my phone limits the time for data collection to an hour or less. I
doubt if either of these tools are appropriate for a course-based
assignment simply because of the range of phones students might possess.

My rough estimates are that I can collect around 250 house numbers in an
hour (either with KeyPadMapper3, a notebook, or a digital audio recorder).
Photo mapping greatly increases the amount of data which can be collected
because facts can be checked. Similarly using FHRS address data means that
trying to record POIs and their addresses can be simplified simply by using
the FHRS data after the survey. In practice most places really require a
second survey to correct omissions or peculiarities which only become
obvious when entering the data.

One other point about metrics such as POI collection: it will greatly
depend on POI density (city centre vs. suburbs) and the current number of
POIs already mapped. It's fundamentally easier to map somewhere de novo
rather than identifying what is missing and what has changed in somewhere
which has already been partially mapped.

In general the longest part of adding data is the building outlines.

There has been some discussion on this topic in the past on this list.

Jerry

On 27 November 2014 at 15:34, Matthijs Melissen <info at matthijsmelissen.nl>
wrote:

> On 27 November 2014 at 16:35, SK53 <sk53.osm at gmail.com> wrote:
> > I have metrics from my own surveys of this type, but would expect a
> shopping
> > street 200 metres long would occupy a student for perhaps a couple of
> hours
> > (certainly my first shopping street, Floral Street in London, took quite
> a
> > long time). In general data collection is much faster than data entry (a
> > ratio of 1:3 or 1:5 for experienced mappers probably more for
> inexperienced
> > mappers).
>
> I think this really depends on the mapper and the method used. My
> collection:entry rate is somewhere between 1:1 and 2:1. In retail
> areas, I collect roughly 80 points of interest per hour, and I can
> enter, very roughly, about 100 points of interest per hour.
>
> I collect data by adding POI as favourites in OsmAND (using both the
> map and Bing layers), and then I export my favourites to JOSM. That
> way, I don't need to type out POI names twice (still a check for typos
> made on mobile is necessary of course), so it goes relatively quick.
>
> Of course, metrics like this are not useful to estimate the time
> students will spend on an assignment. I was taught that at university
> level, students will take roughly eight times as much time to complete
> an exercise as their instructor, and I think in this case the factor
> will even be larger.
>
> -- Matthijs
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-gb/attachments/20141127/87231d2a/attachment-0001.html>


More information about the Talk-GB mailing list