search

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

Add terrain features #7

Closed bensLine closed 8 years ago

bensLine commented 8 years ago

Add features to describe the terrain by using the existing location feature. Possible features used in a tutorial are the proximity to:

bensLine commented 8 years ago

Those features might be provided by group A, follow the discussion at https://github.com/PokemonGoers/PokeData/issues/71

sacdallago commented 8 years ago

Some data you might need to generate yourself. PokeData should be about Pokémon data, not terrain features of the actual world :) it would be also unfair to ask the people on group A to produce this data which is irrelevant for their task ;)

It also introduces you to an important concept in data mining: you need to make feature engineering to discover new data which might be interesting to build a better model, but which is not directly part of the initial test data. This is not to restrain from asking project A to implement new data sources, but please separate between things that are relevant to your project and things that are relevant to everyone/the concept of a Pokémon-centric API

bensLine commented 8 years ago

Yes, of course! It was not meant as a request, we just want to avoid implementing features twice :)

bensLine commented 8 years ago

might be of interest: http://pokemongohub.net/pokemon-go-spawn-points-modeled-open-street-map-data/

jonas-he commented 8 years ago

@bensLine feature is implemented from our side, sorry for the long wait (see https://github.com/PokemonGoers/PokeData/issues/130)

MatthiasBaur commented 8 years ago

Looks nice, thank you!

bensLine commented 8 years ago

PokemonGoers/PokeData#130 was adapted and added in #36 and #37