PokemonGoers / PredictPokemon-2

In this project we will apply machine learning to establish the TLN (Time, Location and Name - that is where pokemons will appear, at what date and time, and which Pokemon will it be) prediction in Pokemon Go.
Apache License 2.0
9 stars 3 forks source link

Create API #44

Closed bensLine closed 8 years ago

bensLine commented 8 years ago

Create a API which team E will use to get predictions for a given location and timestamp.

return results as list

sacdallago commented 8 years ago

I would suggest the programmatical API, not the REST API. Aka: just provide a way for people to import your package var P = require('predictions..'); and then call something like P.predict(lat,lng,timestamp)

gyachdav commented 8 years ago

Additional params should be:

What other params can we offload to the clients?

sacdallago commented 8 years ago

@gyachdav but this should be optional params, right? Theoretically it's trough a call of .predict(..) that these features get added to the data-point to then generate a prediction. Outsourcing this to the caller is shifting the computation to a different place, but at the end of the day it's always gonna be the same process/machine that runs the computation (though, sure: if in the future someone finds a faster API for weather that returns the same params as the one used for the predictions, then they can plug this info in)

goldbergtatyana commented 8 years ago

We found the weather feature not to be relevant for the predictions. Thanks a lot @semioniy for so tirelesly working on the data for this particular feature. @MatthiasBaur will provide an overview in issue 62 of features that will be used in the end.