9 from data_about_players
import Data
11 from knn
import KNNOutputVectorGenerator
14 def __init__( self
, filename
):
15 self
.null
= open('/dev/null','w')
17 s
= "./gnet/gnet_train -l 3 -n 30 -p 30 -e 0.0003 -o gonet.net ./"+filename
19 ret
= subprocess
.call(args
,stdout
=self
.null
)
20 s
= "./gnet/gnet_run gonet.net"
22 self
.p
= subprocess
.Popen(args
, stdin
=subprocess
.PIPE
, stdout
=subprocess
.PIPE
, stderr
=self
.null
)
23 def __call__(self
, vector
):
24 self
.p
.stdin
.write(' '.join([str(a
) for a
in vector
]) + '\n')
25 a
= self
.p
.stdout
.readline()
26 return [ float(num
) for num
in a
.split()]
32 if __name__
== '__main__':
33 main_pat_filename
= Data
.main_pat_filename
43 player_vector
= Data
.questionare_total
44 # players_ignore = [ "Yi Ch'ang-ho 2004-" ]#, "Fujisawa Hideyuki","Yuki Satoshi", "Otake Hideo", "Yi Ch'ang-ho 2005+","Takao Shinji","Hane Naoki","Kobayashi Koichi" ]
45 players_ignore
= [ "Yi Ch'ang-ho 2004-", "Yi Ch'ang-ho 2005+" ]#,"Takao Shinji","Hane Naoki","Kobayashi Koichi" ]
46 players_all
= [ p
for p
in player_vector
.keys() if p
not in players_ignore
]
48 ### Object creating input vector when called
49 print "Creating input vector generator from main pat file:", main_pat_filename
50 i
= InputVectorGenerator(main_pat_filename
, num_features
)
52 # Create list of input vectors
54 for name
in players_all
:
55 input_vectors
+= [i(Data
.pat_files_folder
+ name
)]
57 #print '"%s"'%(players_all[2],)
58 #print input_vectors[2]
60 if len(input_vectors
) == 0:
61 print >>sys
.stderr
, "No reference vectors."
65 # Change this to False, if you do not want to use PCA
68 # Create PCA object, trained on input_vectors
69 print >>sys
.stderr
, "Running PCA."
70 pca
= PCA(input_vectors
, reduce=True)
71 # Perform a PCA on input vectors
72 input_vectors
= pca
.process_list_of_vectors(input_vectors
)
73 # Creates a Composed object that first generates an input vector
74 # and then performs a PCA analysis on it.
77 ### n/4-fold cross validation
78 #bounds = random.sample(range(1,len(players_all)), len(players_all) / 10 )
80 for x
in range(1,len(players_all
)/4):
81 bounds
+= [4*x
for _
in [1] if 4*x
< len(players_all
)]
83 print >>sys
.stderr
, "Pop too small."
88 return [ (x
- 1) / 4.5 - 1.0 for x
in vec
]
90 return [ (x
+ 1) * 4.5 + 1.0 for x
in vec
]
93 return list(2.0*numpy
.random
.random(k
)-1.0)
95 print >>sys
.stderr
, "Running Cross-validation."
97 errs
=[ [] for _
in xrange(len(players_all
)) ]
100 sentinel
=len(players_all
)
102 for _
in xrange(number_runs
):
103 pairs
= zip(players_all
, input_vectors
)
104 random
.shuffle(pairs
)
105 players_all
= [ a
for a
, b
in pairs
]
106 input_vectors
= [ b
for a
, b
in pairs
]
108 for b
in bounds
+[sentinel
]:
109 validation_set
= range(prev
, b
)
110 reference_set
= range(0,prev
) + range(b
,sentinel
)
112 print "Reference set :",
113 for pr
in range(0, prev
):
115 for pr
in validation_set
:
117 for pr
in range(b
, sentinel
):
123 for index
in reference_set
:
124 data
.append( (input_vectors
[index
], norm(player_vector
[players_all
[index
]])) )
127 ### We can enlarge the data set by adding linear combinations of input and output vectors
128 use_lin_combinations
= False
129 if use_lin_combinations
:
130 data
+= Combinator().combine(data
)
132 print_set_to_file(data
,'nn_cross.data')
134 nn
= NeuralNet('nn_cross.data')
135 # Create list of output vectors using weighted kNN algorithm approximating output_vector
136 output_vectors
= [ nn(input_vectors
[index
]) for index
in validation_set
]
139 ### Object creating output vector when called;
141 for index
in reference_set
:
142 ref_dict
[tuple(input_vectors
[index
])] = norm(player_vector
[players_all
[index
]])
145 # best pro InputVectorGenerator rescale=Rescale
146 oknn
= KNNOutputVectorGenerator(ref_dict
, k
=3, weight_param
=0.8)
148 # Create list of output vectors using weighted kNN algorithm approximating output_vector
149 output_vectors
= [ oknn(input_vectors
[index
]) for index
in validation_set
]
150 elif typ
== 'joint_nn_knn':
153 for index
in reference_set
:
154 data
.append( (input_vectors
[index
], norm(player_vector
[players_all
[index
]])) )
155 ref_dict
[tuple(input_vectors
[index
])] = norm(player_vector
[players_all
[index
]])
157 print_set_to_file(data
,'nn_cross.data')
158 nn
= NeuralNet('nn_cross.data')
159 # Create list of output vectors using weighted kNN algorithm approximating output_vector
160 ov_3
= [ nn(input_vectors
[index
]) for index
in validation_set
]
164 oknn
= KNNOutputVectorGenerator(ref_dict
, k
=3, weight_param
=0.8)
165 ov_1
= [ oknn(input_vectors
[index
]) for index
in validation_set
]
167 oknn
= KNNOutputVectorGenerator(ref_dict
, k
=1, weight_param
=0.8)
168 ov_2
= [ oknn(input_vectors
[index
]) for index
in validation_set
]
170 oknn
= KNNOutputVectorGenerator(ref_dict
, k
=1, weight_param
=0.8)
171 ov_4
= [ oknn(input_vectors
[index
]) for index
in validation_set
]
173 output_vectors
= [ [a
[0],b
[1],c
[2],d
[3]] for a
,b
,c
,d
in zip(ov_1
, ov_2
, ov_3
, ov_4
)]
175 output_vectors
= [ rand_vect(4) for index
in validation_set
]
177 output_vectors
= [ revnorm(x
) for x
in output_vectors
]
178 desired_vectors
= [ player_vector
[players_all
[index
]] for index
in validation_set
]
179 #desired_vectors = [ norm(player_vector[players_all[index]]) for index in validation_set ]
182 for vec_set
,text
in [(output_vectors
, "Output: "), (desired_vectors
, "Desired:")]:
190 for num1
, (o
,d
) in zip(validation_set
, zip(output_vectors
, desired_vectors
)):
192 for num
,(x
,y
) in enumerate(zip(o
,d
)):
199 if typ
== 'joint_nn_knn':
200 print "Joint classifier:"
202 print "k-NN classifier:"
204 print "Neural network classifier:"
206 print "Random classifier:"
207 #print "Total square err: %2.3f"%( sum(errs) / number_runs,)
208 mar
= numpy
.array(errs
)
210 print "Mean square err per player: " + "%2.3f ( = sd %2.3f) "%(mean
, sqrt(mean
))
211 mean
= numpy
.array(es
).mean()
212 print "Mean square err per style: " + "%2.3f ( = sd %2.3f) "%(mean
, sqrt(mean
))
213 for num
, style
in enumerate(esps
):
214 mean
= numpy
.array(style
).mean()
215 print "Style %1d : %2.3f ( = sd %2.3f)"%(num
+1, mean
, sqrt(mean
))
216 #print "%2.3f &"%(mean,),
218 #mean = numpy.array(es).mean()
219 #print "%2.3f &"%(mean),
220 #print "%2.3f \\\\\\hline"%(11.776 / mean)
223 #print "Players sorted by mean square error:"
224 #p = zip([numpy.array(errs[p]).mean() for p in xrange(len(players_all)) ], players_all)
227 # print "%2.3f %s"%(err,name)
228 # #print "%s"%(name,)