-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathnai_m_d.py
245 lines (210 loc) · 12.9 KB
/
nai_m_d.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
import time; start_time = time.time()
import numpy as np
import pandas as pd
from sklearn import ensemble
import xgboost as xgb
from sklearn.metrics import log_loss, make_scorer
from sklearn.grid_search import GridSearchCV
from sklearn.cross_validation import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.neighbors import KNeighborsClassifier
from sklearn.linear_model import SGDClassifier
from sklearn.mixture import GMM
from sklearn.discriminant_analysis import LinearDiscriminantAnalysis, QuadraticDiscriminantAnalysis
from sklearn.svm import SVC
from sklearn.linear_model import RidgeClassifier, Ridge
#from enn import ENN
from scipy.spatial.distance import euclidean, mahalanobis
from sklearn.linear_model import PassiveAggressiveClassifier
from sklearn.linear_model import Perceptron
from sklearn.naive_bayes import MultinomialNB
from lightning.classification import CDClassifier, FistaClassifier, SAGAClassifier, LinearSVC
from sklearn.neighbors.nearest_centroid import NearestCentroid
#from pyearth import Earth
from sklearn.cluster import k_means
from elm import ELMClassifier, ELMRegressor, GenELMClassifier, GenELMRegressor
from random_layer import RandomLayer, MLPRandomLayer, RBFRandomLayer, GRBFRandomLayer
import random; random.seed(2016)
import sys
d=4
#d=int(sys.argv[2]) #d external parameter
print("file: "+str(d)+"\n")
train = pd.read_csv('input/numerai_training_data_'+str(d)+'.csv', encoding="utf-8-sig")
test = pd.read_csv('input/numerai_tournament_data_'+str(d)+'.csv', encoding="utf-8-sig")
num_train = train.shape[0]
y_train = train['target']
train = train.drop(['target'],axis=1)
id_test = test['t_id']
def fill_nan_null(val):
ret_fill_nan_null = 0.0
if val == True:
ret_fill_nan_null = 1.0
return ret_fill_nan_null
df_all = pd.concat((train, test), axis=0, ignore_index=True)
df_all['null_count'] = df_all.isnull().sum(axis=1).tolist()
df_all_temp = df_all['t_id']
df_all = df_all.drop(['t_id'],axis=1)
df_data_types = df_all.dtypes[:] #{'object':0,'int64':0,'float64':0,'datetime64':0}
d_col_drops = []
df_all = pd.concat([df_all, df_all_temp], axis=1)
print(len(df_all), len(df_all.columns))
train = train.drop(d_col_drops,axis=1)
test = test.drop(d_col_drops,axis=1)
def flog_loss(ground_truth, predictions):
flog_loss_ = log_loss(ground_truth, predictions) #, eps=1e-15, normalize=True, sample_weight=None)
return flog_loss_
LL = make_scorer(flog_loss, greater_is_better=False)
#m=int(sys.argv[1]) #m external parameter
m=1
print("m: "+str(m)+"\n")
#mf - max features
if m==1:
g={'ne':150,'md':6,'mf':21+d,'rs':2016} #change to gne':500,'md':40,'mf':60,'rs':2016}
elif m==2:
g={'ne':500,'md':40,'mf':21+d,'rs':2016}
elif m==3:
g={'ne':300,'md':6,'mf':21+d,'rs':2016}
elif m==4:
g={'ne':150,'md':5,'mf':21+d,'rs':2016}
elif m==5:
g={'ne':300,'md':4,'mf':21+d,'rs':2016}
elif m==6:
g={'ne':300,'md':8,'mf':21+d,'rs':2016}
elif m==7:
g={'ne':150,'md':10,'mf':21+d,'rs':2016}
elif m==8:
g={'ne':400,'md':8,'mf':15+d,'rs':2016}
elif m==9:
g={'ne':500,'md':9,'mf':15+d,'rs':2016}
elif m==10:
g={'ne':500,'md':10,'mf':20+d,'rs':2016}
elif m==11:
g={'ne':400,'md':7,'mf':21+d,'rs':2016}
#logistic regression
lg1c = LogisticRegression(solver = 'liblinear', penalty = 'l1', C = 1.0, n_jobs = -1)
lg2c = LogisticRegression(solver = 'liblinear', penalty = 'l1', C = .01, n_jobs = -1)
lg3c = LogisticRegression(solver = 'liblinear', penalty = 'l1', C = 100, n_jobs = -1)
lg4c = LogisticRegression(solver = 'liblinear', penalty = 'l2', C = 1.0, n_jobs = -1) #C=1.0 default
lg5c = LogisticRegression(solver = 'liblinear', penalty = 'l2', C = .01, n_jobs = -1)
lg6c = LogisticRegression(solver = 'liblinear', penalty = 'l2', C = 100, n_jobs = -1)
#
#lg7c = LogisticRegression(penalty='l2', C=1.0, fit_intercept=True, solver='sag', max_iter=10000, n_jobs=-1)
lg7c = LogisticRegression(solver = 'liblinear', penalty = 'l1', C = 10, n_jobs = -1)
lg8c = LogisticRegression(solver = 'liblinear', penalty = 'l2', C = 10, n_jobs = -1)
#ridge
ridge1 = Ridge(alpha=1.0, copy_X=True, fit_intercept=True, max_iter=None,normalize=False, random_state=None, solver='auto', tol=0.001)
ridge2 = Ridge(alpha=1.0, copy_X=True, fit_intercept=True, max_iter=None,normalize=False, random_state=None, solver='svd', tol=0.001)
ridge3 = Ridge(alpha=1.0, copy_X=True, fit_intercept=True, max_iter=None,normalize=False, random_state=None, solver='cholesky', tol=0.001)
ridge4 = Ridge(alpha=1.0, copy_X=True, fit_intercept=True, max_iter=None,normalize=False, random_state=None, solver='lsqr', tol=0.001)
ridge5 = Ridge(alpha=1.0, copy_X=True, fit_intercept=True, max_iter=None,normalize=False, random_state=None, solver='sparse_cg', tol=0.001)
ridge6 = Ridge(alpha=1.0, copy_X=True, fit_intercept=True, max_iter=None,normalize=False, random_state=None, solver='sag', tol=0.001)
#ridge7 = Ridge(alpha=0.1, copy_X=True, fit_intercept=True, max_iter=10000,normalize=False, random_state=None, solver='auto', tol=0.001)
#lda,qda
ldac = LinearDiscriminantAnalysis()
qdac = QuadraticDiscriminantAnalysis()
#knn
knn1c = KNeighborsClassifier(n_neighbors=25, p = 1, weights = 'distance',n_jobs=-1)
knn2c = KNeighborsClassifier(n_neighbors=25, p = 2, weights = 'distance',n_jobs=-1 )
knn3c = KNeighborsClassifier(n_neighbors=25, metric = 'chebyshev', weights = 'distance',n_jobs=-1)
nc1 = NearestCentroid()
#semi-supervised
#LabelPropagation
#LabelSpreading
#perceptron
per1 = Perceptron(n_jobs=-1, n_iter=10)
per2 = Perceptron(penalty='l2', n_jobs=-1, n_iter=10)
per3 = Perceptron(penalty='l1', n_jobs=-1, n_iter=10)
per4 = Perceptron(penalty='elasticnet', n_jobs=-1, n_iter=10)
#passive aggresive
passaggr = PassiveAggressiveClassifier(n_jobs=-1, n_iter=10)
#lightning
CD1_c = CDClassifier(loss='log', penalty='l2', multiclass=False, C=1.0, alpha=1.0, max_iter=50, tol=0.001, termination='violation_sum', shrinking=True, max_steps='auto', sigma=0.01, beta=0.5, warm_start=False, debiasing=False, Cd=1.0, warm_debiasing=False, selection='cyclic', permute=True, callback=None, n_calls=100, random_state=None, verbose=1)
CD2_c = CDClassifier(loss='modified_huber', penalty='l2', multiclass=False, C=1.0, alpha=1.0, max_iter=50, tol=0.001, termination='violation_sum', shrinking=True, max_steps='auto', sigma=0.01, beta=0.5, warm_start=False, debiasing=False, Cd=1.0, warm_debiasing=False, selection='cyclic', permute=True, callback=None, n_calls=100, random_state=None, verbose=1)
SAGA_c = SAGAClassifier(eta='auto', alpha=1.0, beta=0.0, loss='modified_huber', penalty=None, gamma=1.0, max_iter=100, n_inner=1.0, tol=0.001, verbose=0, callback=None, random_state=None)
#Fista1 = FistaClassifier(C=1.0, alpha=1.0, loss='log', penalty='l2', multiclass=False, max_iter=100, max_steps=30, eta=2.0, sigma=1e-05, callback=None, verbose=1)
#Fista2_c = FistaClassifier(C=1.0, alpha=1.0, loss='modified_huber', penalty='l1', multiclass=False, max_iter=100, max_steps=30, eta=2.0, sigma=1e-05, callback=None, verbose=0)
#SAGA_c = SAGAClassifier(eta='auto', alpha=1.0, beta=0.0, loss='modified_huber', penalty=None, gamma=1.0, max_iter=100, n_inner=1.0, tol=0.001, verbose=0, callback=None, random_state=None)
#LinSVC_c = LinearSVC(C=1.0, loss='hinge', criterion='accuracy', max_iter=1000, tol=0.001, permute=True, shrinking=True, warm_start=False, random_state=None, callback=None, n_calls=100, verbose=0)
#earth
#earth_c = Earth()
elm1 = ELMClassifier(n_hidden=500, activation_func='multiquadric')
elm2 = ELMRegressor(random_state=0, activation_func='gaussian', alpha=0.1)
elm3 = ELMClassifier(n_hidden=1000, activation_func='gaussian', alpha=0.1, random_state=0)
elm4 = ELMClassifier(n_hidden=500, activation_func='hardlim', alpha=0.1, random_state=0)
elm5 = ELMRegressor(random_state=0, activation_func='tanh', alpha=0.1)
elm6 = ELMRegressor(random_state=0, activation_func='sine', alpha=0.1)
elm7 = ELMRegressor(random_state=0, activation_func='tribas', alpha=0.1)
elm8 = ELMRegressor(random_state=0, activation_func='inv_tribas', alpha=0.1)
elm9 = ELMRegressor(random_state=0, activation_func='sigmoid', alpha=0.1)
elm10 = ELMRegressor(random_state=0, activation_func='hardlim', alpha=0.1)
elm11 = ELMRegressor(random_state=0, activation_func='softlim', alpha=0.1)
elm12 = ELMRegressor(random_state=0, activation_func='multiquadric', alpha=0.1)
elm13 = ELMRegressor(random_state=0, activation_func='inv_multiquadric', alpha=0.1)
#'tanh', 'sine', 'tribas', 'inv_tribase', 'sigmoid','hardlim', 'softlim', 'gaussian', 'multiquadric', 'inv_multiquadric'
#orig
etc = ensemble.ExtraTreesClassifier(n_estimators=g['ne'], max_depth=g['md'], max_features=g['mf'], random_state=g['rs'], criterion='entropy', min_samples_split= 4, min_samples_leaf= 2, verbose = 0, n_jobs =-1)
etr = ensemble.ExtraTreesRegressor(n_estimators=g['ne'], max_depth=g['md'], max_features=g['mf'], random_state=g['rs'], min_samples_split= 4, min_samples_leaf= 2, verbose = 0, n_jobs =-1)
rfc = ensemble.RandomForestClassifier(n_estimators=g['ne'], max_depth=g['md'], max_features=g['mf'], random_state=g['rs'], criterion='entropy', min_samples_split= 4, min_samples_leaf= 2, verbose = 0, n_jobs =-1)
rfr = ensemble.RandomForestRegressor(n_estimators=g['ne'], max_depth=g['md'], max_features=g['mf'], random_state=g['rs'], min_samples_split= 4, min_samples_leaf= 2, verbose = 0, n_jobs =-1)
xgr = xgb.XGBRegressor(n_estimators=g['ne'], max_depth=g['md'], seed=g['rs'], missing=np.nan, learning_rate=0.02, subsample=0.9, colsample_bytree=0.85, objective='reg:linear')
xgc = xgb.XGBClassifier(n_estimators=g['ne'], max_depth=g['md'], seed=g['rs'], missing=np.nan, learning_rate=0.02, subsample=0.9, colsample_bytree=0.85, objective='binary:logistic') #try 'binary:logitraw'
#clf = {'elm1':elm1,'elm2':elm2,'elm3':elm3,'elm4':elm4,'elm5':elm5,'elm6':elm6,
# 'elm7':elm7,'elm8':elm8,'elm9':elm9,'elm10':elm10,'elm11':elm11,'elm12':elm12,'elm13':elm13}
#clf = {'nc1':nc1}
#clf = {'SAGA_c':SAGA_c}#,'Fista2_c':Fista2_c}
#clf = {'earth_c':earth_c}
#clf = {'CD1_c':CD1_c,'CD2_c':CD2_c}
#clf = {'ridge1':ridge1,'ridge2':ridge2,'ridge3':ridge3,'ridge4':ridge4,'ridge5':ridge5,'ridge6':ridge6}
#clf = {'per1':per1, 'per2':per2,'per3':per3,'per4':per4,'passagr':passaggr}
#clf = {'knn1c':knn1c,'knn2c':knn2c,'knn3c':knn3c}
#clf = {'ldac':ldac,'qdac':qdac}
#final set of classifiers:
clf = {'lg1c':lg1c,'lg2c':lg2c,'lg3c':lg3c,'lg4c':lg4c,'lg5c':lg5c,'lg6c':lg6c,'lg7c':lg7c,'lg8c':lg8c}
# 'etc':etc, 'etr':etr, 'rfc':rfc, 'rfr':rfr, 'xgr':xgr, 'xgc':xgc}
y_pred=[]
best_score = 0.0
id_results = id_test[:]
for c in clf:
if c[:1] != "x": #not xgb
model = GridSearchCV(estimator=clf[c], param_grid={}, n_jobs =-1, cv=2, verbose=0, scoring=LL)
model.fit(train, y_train.values)
if c[-1:] != "c": #not classifier
y_pred = model.predict(test)
if c[:3] == "elm":
print("Ensemble Model: ", c, " Best CV score: ", model.best_score_)#, " Time: ", round(((time.time() - start_time)/60),2))
else:
print("Ensemble Model: ", c, " Best CV score: ", model.best_score_, " Time: ", round(((time.time() - start_time)/60),2))
else: #classifier
best_score = (log_loss(y_train.values, model.predict_proba(train)))*-1
y_pred = model.predict_proba(test)[:,1]
print("Ensemble Model: ", c, " Best CV score: ", best_score, " Time: ", round(((time.time() - start_time)/60),2))
else: #xgb
X_fit, X_eval, y_fit, y_eval= train_test_split(train, y_train, test_size=0.35, train_size=0.65, random_state=g['rs'])
model = clf[c]
model.fit(X_fit, y_fit.values, early_stopping_rounds=20, eval_metric="logloss", eval_set=[(X_eval, y_eval)], verbose=0)
if c == "xgr": #xgb regressor
best_score = (log_loss(y_train.values, model.predict(train)))*-1
y_pred = model.predict(test)
else: #xgb classifier
best_score = (log_loss(y_train.values, model.predict_proba(train)))*-1
y_pred = model.predict_proba(test)[:,1]
print("Ensemble Model: ", c, " Best CV score: ", best_score, " Time: ", round(((time.time() - start_time)/60),2))
for i in range(len(y_pred)):
if y_pred[i]<0.0:
y_pred[i] = 0.0
if y_pred[i]>1.0:
y_pred[i] = 1.0
df_in = pd.DataFrame({"t_id": id_test, c: y_pred})
id_results = pd.concat([id_results, df_in[c]], axis=1)
id_results['avg'] = id_results.drop('t_id', axis=1).apply(np.average, axis=1)
id_results['min'] = id_results.drop('t_id', axis=1).apply(min, axis=1)
id_results['max'] = id_results.drop('t_id', axis=1).apply(max, axis=1)
id_results['diff'] = id_results['max'] - id_results['min']
#for i in range(10):
# print(i, len(id_results[id_results['diff']>(i/10)]))
id_results.to_csv("output/results_analysis_"+str(d)+"_"+str(m)+".csv", index=False)
ds = id_results[['t_id','avg']]
ds.columns = ["t_id","probability"]
ds.to_csv("output/submission_"+str(d)+"_"+str(m)+".csv",index=False)
#import xgbfir
#xgbfir.saveXgbFI(xgc, OutputXlsxFile = "ouput/xgbfir_"+str(d)+"_"+str(m)+".xlsx")