Just did a demo video for this to show off to the Numenta team... might as well share with all of you: https://youtu.be/0lFFI86sa2Q
Has anyone used it yet? It makes me want to get a lot more data into RV... Regards, --------- Matt Taylor OS Community Flag-Bearer Numenta On Tue, Oct 27, 2015 at 12:40 PM, Matthew Taylor <[email protected]> wrote: > Hello NuPIC, > > I would like to introduce a project I've been working on called menorah: > > https://github.com/nupic-community/menorah > > Menorah makes it really easy to run data directly from River View into > NuPIC. With it, you can swarm over RV data, then turn around and > create a model from the best model params from the swarm, immediately > passing the live River View data into the model. > > It will also work with multiple streams of data, so you can specify > many data feeds to be passed into your model. Menorah will pull all > data from River View, format it into the proper CSV format required by > NuPIC, and run a swarm for a specified predicted field. > > Because you specify a working directory for each menorah instance, all > the NuPIC artifacts from your swarm will be stored there to be easily > re-used later to run your models. > > Have a look at a quick example from the README: > > ======================================== > from datetime import datetime > > from menorah import Menorah > > sources = [ > ["chicago-311", "Tree Debris", "aggregate=1 day"], > ["chicago-beach-weather", "Foster Weather Station", "humidity"], > ["chicago-beach-weather", "Foster Weather Station", "interval_rain"], > ["chicago-beach-water-quality", "Osterman Beach", "wave_height"], > ] > > menorah = Menorah( > sources, > "work/example5-multifield-aggregated", > since=datetime(2015, 5, 20) > ) > > menorah.swarm(swarmParams={"swarmSize":"large"}) > menorah.runModel(plot=True) > ======================================== > > These few lines of code will load 4 River View streams into memory and > swarm over all that data with the "Tree Debris" as the predicted > field. Immediately after the swarm, the "runModel()" call will reset > the data streams and feed all the same data into a new model created > using the best model params returned by the swarm. > > This could be really useful for some of you having trouble getting > your data into the right format. If you can write a River View parser > for your data, you can make NuPIC predict it! > > Hope this is helpful, > > --------- > Matt Taylor > OS Community Flag-Bearer > Numenta
