• 34 comments

  •   •   over 8 years ago

    My interpretation is that the service doesn't exist. There is 'a' service that exists, but seems extremely incomplete for sure. I actually wonder how accurate is the data in there. Is it near real time?

    @John, the point was not that the homeless would update the number of beds. My point is that the solution should probably include a mechanism for somebody (shelters) to update that info.
    Again, I think that's the whole point of this challenge.

    If the service that already exists, what were you planning to do, add some CSS to make it look pretty?

  •   •   over 8 years ago

    I would say make it work better and inform the public more completely, not just to make it pretty, but your point is understood.

    It would be nice to get confirmation from someone with solid information...

  •   •   over 8 years ago

    @John, I agree, confirmation of what is expected would be nice...

  •   •   over 8 years ago

    So... I really think we need to gather our own data for beds available.

    Even if we were to parse the data at https://hmis.njhmfaserv.org/ just look at that data now, and look again tomorrow.... yeah, I don't know if that data is ever updated, not quite real time.

  • Manager   •   over 8 years ago

    This is an important discussion. Jean-Philippe is correct to note that current data sets regarding bed availability are very limited, but do serve as a baseline to determine usefulness of a submission. We are working on getting a standard regarding "near real time" finalized so that its clearer what folks need to focus on in order to be competitive as a Finalist.

    Please keep an eye on the Updates section of the Project REACH site. We'll post information there as soon as we have it.

    Best,
    VAi2 Team

  • Manager   •   over 8 years ago

    Achieving the Near-Real Time Bed Availability Criteria:

    The New Jersey HMIS team has agreed to begin refreshing its data about shelter bed availability every six hours instead of once a day.

    Of course six hours is not exactly “real time”, using this dataset does fulfill the requirement for “near-real time” bed availability, and thereby establishes a quantitative definition of “near”.

    Providing near-real time bed availability is considered the “floor” and not the “ceiling” for the following Project REACH evaluation criteria; we’d love to see clever solutions that could make it even faster:

    “Creates a mechanism that allows for simple aggregation of near or real time bed availability at shelters local to JBJ Soul Kitchen, which is located at: 207 Monmouth St., Red Bank, New Jersey 07701.”

    Instructions for downloading the new Near-Real Time Bed Availability data set:

    1. Visit https://hmis.njhmfaserv.org/lst/?p=pol

    2. Under the heading “Developer,” locate the file named "Project Reach Datafile"

    3. Download the file named "bedregistrYYYYMMDDhhmmss"

    We’d like to express a HUGE thank you to our HUD colleagues and the New Jersey HMIS team.

  •   •   over 8 years ago

    Any chance they could add a csv version of the Project Reach Datafile? And maybe the address for each location?

  • Manager   •   over 8 years ago

    @Casey T: You read our mind (and others'). The good folks at HMIS will start issuing the file with address location included.

    As for .csv format, we'll check on it and get back.

    Best,
    VAi2 Team

    UPDATE: The data set is now available is .csv!

  •   •   over 8 years ago

    If the Foothold developers could do us one more favor and add a column for the HUD HMIS program type that would be ideal. Currently, the table provided doesn't show this which will result in confusion if someone is looking for a bed for the night and permanent/long term housing options pop up. CSP is one example of an agency that has programs that are purely long-term supportive housing for clients with mental illness. It is OK to capture details on this bed inventory, as each and every bed is a key component to the community response to the issue of homelessness, but it should be clear for end-users of the app that they should not be calling up these program staff for people that are in immediate need for emergency shelter (program type = 1). For those that are unaware of the program types they are listed on page 29 of the HUD HMIS Data Standards found at http://www.hmis.info/ClassicAsp/documents/Final%20HMIS%20Data%20Standards-Revised%20_3.pdf. Also, if we are ever to get this data in real-time then it would be ideal if they could pass along the Organization and Program IDs next to the Org and Program names. This way, we could eventually adopt a model whereby we adjust the bed availability each time a bed or unit is consumed using web services and data provided in either the HUD XML or HUD CSV format.

    Program types of 2=transitional and 3=permanent could eventually utilize a centralized reservation system similar to what hotel chains might use as these types of housing do not turn over as frequently as emergency shelter beds. This way, if a case manager is working with a family they can not only see what is open today but can see what is open two weeks from now and place the family in accordance to both availability as well as other factors such as proximity to the children's schools, community supports, jobs, etc.. Typically if a family has an eviction notice in hand they will ride out that eviction just in case they can do anything to stay but they will want to have a place to move to on their eviction date if all else fails. If a case manager re-houses a family into another housing option prematurely, just to secure them an open unit, then they are decreasing the supply of shelter for those that have a more immediate need and it introduces the family into homelessness when there may have been ways it still could have been avoided. Reservation systems can help prevent this.

    A couple of additional points to consider would be to track beds and units separately as this allows for a quick delineation between family and individual programs. Last suggestion would be to consider adding columns for reserved and offline beds and units. With this, we could calculate the total number of open beds so that column could be removed altogether. If you look at the chart provided it is obvious that this data is somewhere as if you subtract the total # of beds occupied from the total number of beds it does not always result in the total number of open beds. Considering that none of these points have come up in the forums suggests that many that are participating are going to be using something simple such as Google Places to identify service locations. This will not work for this challenge as there needs to be mechanisms in place to "provide real- or near real-time information about bed availability" which will require the proper database hooks that Google Places and others do not support natively.

Comments are closed.