Skip to content

Instantly share code, notes, and snippets.

@rimms

rimms/bin.json

Created September 4, 2014 05:04
Show Gist options
  • Save rimms/b0a192112df745adfa88 to your computer and use it in GitHub Desktop.
Save rimms/b0a192112df745adfa88 to your computer and use it in GitHub Desktop.
Download ZIP
NN_with_idf
Raw
bin.json
{
"converter" : {
"string_filter_types": {},
"string_filter_rules":[],
"num_filter_types": {},
"num_filter_rules": [],
"string_types": {
"unigram": { "method": "ngram", "char_num": "1" }
},
"string_rules":[
{"key" : "*", "type" : "unigram", "sample_weight":"bin", "global_weight" : "bin"}
],
"num_types": {},
"num_rules": []
},
"parameter" : {
"hash_num" : 64
},
"method": "euclid_lsh"
}
Raw
idf.json
{
"converter" : {
"string_filter_types": {},
"string_filter_rules":[],
"num_filter_types": {},
"num_filter_rules": [],
"string_types": {
"unigram": { "method": "ngram", "char_num": "1" }
},
"string_rules":[
{"key" : "*", "type" : "unigram", "sample_weight":"bin", "global_weight" : "idf"}
],
"num_types": {},
"num_rules": []
},
"parameter" : {
"hash_num" : 64
},
"method": "euclid_lsh"
}
Raw
result
$ jubanearest_neighbor -f bin.json &
$ ./nn_shogn.py
[id_with_score{id: 徳川12, score: 1.27169287205}, id_with_score{id: 足利10, score: 1.57138991356}, id_with_score{id: 足利12, score: 1.57138991356}, id_with_score{id: 北条11, score: 1.68489217758}, id_with_score{id: 足利02, score: 1.79433512688}, id_with_score{id: 足利07, score: 1.79433512688}, id_with_score{id: 北条06, score: 1.79433512688}, id_with_score{id: 徳川02, score: 1.84745168686}, id_with_score{id: 足利08, score: 1.84745168686}, id_with_score{id: 北条09, score: 1.84745168686}]
徳川02 慶喜
[id_with_score{id: 足利03, score: 1.27169287205}, id_with_score{id: 足利13, score: 1.27169287205}, id_with_score{id: 足利14, score: 1.27169287205}, id_with_score{id: 足利15, score: 1.33331120014}, id_with_score{id: 足利04, score: 1.39412653446}, id_with_score{id: 足利06, score: 1.39412653446}, id_with_score{id: 足利09, score: 1.39412653446}, id_with_score{id: 足利11, score: 1.39412653446}, id_with_score{id: 足利07, score: 1.45410203934}, id_with_score{id: 徳川09, score: 1.57138991356}]
徳川09 義昭
[id_with_score{id: 北条14, score: 1.01793670654}, id_with_score{id: 北条12, score: 1.20930838585}, id_with_score{id: 北条03, score: 1.27169287205}, id_with_score{id: 北条08, score: 1.27169287205}, id_with_score{id: 北条09, score: 1.27169287205}, id_with_score{id: 北条10, score: 1.27169287205}, id_with_score{id: 北条13, score: 1.27169287205}, id_with_score{id: 北条01, score: 1.33331120014}, id_with_score{id: 北条02, score: 1.33331120014}, id_with_score{id: 北条04, score: 1.33331120014}]
北条01 守時
$ jubanearest_neighbor -f idf.json &
$ ./nn_shogn.py
[id_with_score{id: 徳川01, score: 0.0}, id_with_score{id: 徳川02, score: 0.0}, id_with_score{id: 徳川03, score: 0.0}, id_with_score{id: 徳川04, score: 0.0}, id_with_score{id: 徳川05, score: 0.0}, id_with_score{id: 徳川06, score: 0.0}, id_with_score{id: 徳川07, score: 0.0}, id_with_score{id: 徳川08, score: 0.0}, id_with_score{id: 徳川09, score: 0.0}, id_with_score{id: 徳川10, score: 0.0}]
徳川01 慶喜
[id_with_score{id: 徳川01, score: 0.0}, id_with_score{id: 徳川02, score: 0.0}, id_with_score{id: 徳川03, score: 0.0}, id_with_score{id: 徳川04, score: 0.0}, id_with_score{id: 徳川05, score: 0.0}, id_with_score{id: 徳川06, score: 0.0}, id_with_score{id: 徳川07, score: 0.0}, id_with_score{id: 徳川08, score: 0.0}, id_with_score{id: 徳川09, score: 0.0}, id_with_score{id: 徳川10, score: 0.0}]
徳川01 義昭
[id_with_score{id: 徳川01, score: 0.0}, id_with_score{id: 徳川02, score: 0.0}, id_with_score{id: 徳川03, score: 0.0}, id_with_score{id: 徳川04, score: 0.0}, id_with_score{id: 徳川05, score: 0.0}, id_with_score{id: 徳川06, score: 0.0}, id_with_score{id: 徳川07, score: 0.0}, id_with_score{id: 徳川08, score: 0.0}, id_with_score{id: 徳川09, score: 0.0}, id_with_score{id: 徳川10, score: 0.0}]
徳川01 守時
Raw
shogun_nn.py
#!/usr/bin/env python
# coding: utf-8
host = '127.0.0.1'
port = 9199
name = 'test'
import sys
import json
import random
import jubatus
from jubatus.common import Datum
def train(client):
client.set_row(u'徳川01', Datum({'name': u'家康'}))
client.set_row(u'徳川02', Datum({'name': u'秀忠'}))
client.set_row(u'徳川03', Datum({'name': u'家光'}))
client.set_row(u'徳川04', Datum({'name': u'家綱'}))
client.set_row(u'徳川05', Datum({'name': u'綱吉'}))
client.set_row(u'徳川06', Datum({'name': u'家宣'}))
client.set_row(u'徳川07', Datum({'name': u'家継'}))
client.set_row(u'徳川08', Datum({'name': u'吉宗'}))
client.set_row(u'徳川09', Datum({'name': u'家重'}))
client.set_row(u'徳川10', Datum({'name': u'家治'}))
client.set_row(u'徳川11', Datum({'name': u'家斉'}))
client.set_row(u'徳川12', Datum({'name': u'家慶'}))
client.set_row(u'徳川13', Datum({'name': u'家定'}))
client.set_row(u'徳川14', Datum({'name': u'家茂'}))
client.set_row(u'足利01', Datum({'name': u'尊氏'}))
client.set_row(u'足利02', Datum({'name': u'義詮'}))
client.set_row(u'足利03', Datum({'name': u'義満'}))
client.set_row(u'足利04', Datum({'name': u'義持'}))
client.set_row(u'足利05', Datum({'name': u'義量'}))
client.set_row(u'足利06', Datum({'name': u'義教'}))
client.set_row(u'足利07', Datum({'name': u'義勝'}))
client.set_row(u'足利08', Datum({'name': u'義政'}))
client.set_row(u'足利09', Datum({'name': u'義尚'}))
client.set_row(u'足利10', Datum({'name': u'義稙'}))
client.set_row(u'足利11', Datum({'name': u'義澄'}))
client.set_row(u'足利12', Datum({'name': u'義稙'}))
client.set_row(u'足利13', Datum({'name': u'義晴'}))
client.set_row(u'足利14', Datum({'name': u'義輝'}))
client.set_row(u'足利15', Datum({'name': u'義栄'}))
client.set_row(u'北条01', Datum({'name': u'時政'}))
client.set_row(u'北条02', Datum({'name': u'義時'}))
client.set_row(u'北条03', Datum({'name': u'泰時'}))
client.set_row(u'北条04', Datum({'name': u'経時'}))
client.set_row(u'北条05', Datum({'name': u'時頼'}))
client.set_row(u'北条06', Datum({'name': u'長時'}))
client.set_row(u'北条07', Datum({'name': u'政村'}))
client.set_row(u'北条08', Datum({'name': u'時宗'}))
client.set_row(u'北条09', Datum({'name': u'貞時'}))
client.set_row(u'北条10', Datum({'name': u'師時'}))
client.set_row(u'北条11', Datum({'name': u'宗宣'}))
client.set_row(u'北条12', Datum({'name': u'煕時'}))
client.set_row(u'北条13', Datum({'name': u'基時'}))
client.set_row(u'北条14', Datum({'name': u'高時'}))
client.set_row(u'北条15', Datum({'name': u'貞顕'}))
def predict(client):
# predict the last shogun
data = [
Datum({'name': u'慶喜'}),
Datum({'name': u'義昭'}),
Datum({'name': u'守時'}),
]
for d in data:
res = client.neighbor_row_from_datum(d, 10)
print res
# get the predicted shogun name
sys.stdout.write(max(res, key = lambda x: x.score).id)
sys.stdout.write(' ')
sys.stdout.write(d.string_values[0][1].encode('utf-8'))
sys.stdout.write('\n')
if __name__ == '__main__':
# connect to the jubatus
client = jubatus.NearestNeighbor(host, port, name)
# run example
train(client)
predict(client)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment