Cool! And then build it as a generic pipeline, to make riverview look upstream the river as well ;)
I would love to work on this, but I'm flooded with work here atm :( too bad timing with the challenge as well :'( But definitely a cool and useful use case! Best, Pascal ____________________________ BE THE CHANGE YOU WANT TO SEE IN THE WORLD ... > On 02 Oct 2015, at 18:30, Matthew Taylor <[email protected]> wrote: > > Hello NuPIC, > > I have been working on a new River View feature [1] and thinking of putting > together another example HTM application (for NuPIC) that uses more than one > scalar input field to make a multistep prediction. For example, it is pretty > obvious if you look into this aggregated data [2] that an event occurred in > early August that caused many people to call the city of Chicago to request > tree debris removal. This should correlate directly with existing Chicago > wind speed and precipitation data [3] (spoiler: it was a tornado [4]). > > Would something like this be an informative sample application? One that > shows how to swarm over multiple input fields to find those that contribute > to the best prediction of another field using live streaming data from River > View? > > Please speak up if you are interested, especially if you think you might want > to do something like this for your HTM Challenge submission. ;) > > [1] > https://github.com/numenta/nupic/wiki/River-View-and-HTM#aggregated-geospatial-data > [2] > http://data.numenta.org/chicago-311/Tree%20Debris/data.html?aggregate=1%20day&since=1435854390 > [3] > http://data.numenta.org/chicago-beach-weather/Foster%20Weather%20Station/data.html?limit=5000 > [4] http://www.weather.gov/lot/2August2015 > > Regards, > --------- > Matt Taylor > OS Community Flag-Bearer > Numenta
