-
Notifications
You must be signed in to change notification settings - Fork 7
/
bike_assign_master_cfg.py
401 lines (341 loc) · 15 KB
/
bike_assign_master_cfg.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
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
import os, time, string
from UserDict import UserDict
from math import sqrt, exp
from numpy import *
import multiprocessing, random
import bikeAssign.misc as rm_misc
from bikeAssign.choice_set.beta_unif_randomizer import BetaUnifRandomizer
class BikeAssignMasterCfg(UserDict):
"""compile configuration data"""
def __init__(self, changes={}):
UserDict.__init__(self)
self.network_config=NetworkConfig()
self.outer_network_config=OuterNetworkConfig()
self.output_config=OutputConfig()
self.choice_set_config=ChoiceSetConfig()
self.assign_config=AssignConfig()
#location of travel data in matsim format
self['travel_dir']=r"X:\Projects\BikeModel\data\bike_model\input\travel\2010_04_18"
#directory with trip start times in hours on 24 hr clock
self['time_file']=r"X:\Projects\BikeModel\data\bike_model\input\context\2010_04_18\time.csv"
#to holdback a sample of observations from estimation to use for validation
self['use_holdback_sample']=True
self['holdback_rate']=0.10
self['holdback_from_file']=True
self['holdback_file']=r"X:\Projects\BikeModel\data\bike_model\input\holdback\Holdback.csv"
self['n_processes']=multiprocessing.cpu_count()
print "Setting n_processes to %d" % (self['n_processes'])
"""deprecated"""
# rules to lookup time dependent variables in network data { variable in choice_set_config
# [ ( (time lower bound, time upper bound) , variable to lookup), ... ,
# ('else', variable to lookup if time outside preceeding bounds) ]}
# else condition must be last for each rule
self['time_dependent_relation']={'V':[((3,6),'V_EA'),((6,9),'V_AM'),((9,15.5),'V_MD'),((15.5,18.5),'V_PM'),('else','V_EV')]}
#this gives the location of travel data with chosen routes that are hard to reproduce, used in route_model.evaluate_parameters
self['imperfect_file']=r"X:\Projects\BikeModel\data\bike_model\input\optim\imperfect.csv"
for key in changes:
self[key]=changes[key]
class NetworkConfig(UserDict):
"""store network configuration data"""
def __init__(self, changes={}):
UserDict.__init__(self)
self['data_dir']=r"bike_model_input"
self['link_file']=os.path.join(self['data_dir'],'links.csv')
self['node_file']=os.path.join(self['data_dir'],'nodes.csv')
self['dist_var']='DISTANCE'
self['dist_scl']=1/5280 #rescales with node distance x dist_scl= link distance
self['max_centroid']=2454
self['exclude_group']={'FT':('in',[1,2,101,102]),'MTYPE_NUM':('==',0)}
self['use_dual']=True
self['perform_transformation']=True
self['create_ww_links']=True
self['ww_exist_alias']=('ONEWAY','WRONG_WAY')
self['ww_change']={'FT':('+',100),'BIKE_CLASS':('*',0),'PER_RISE':('*',-1)}
self['variable_transforms']={
'MTYPE_NUM' :('MTYPE', {'SF':1,'MTC':0} , "" ),
'B0' :('BIKE_CLASS', {0:1,1:0,2:0,3:0} , "int" ),
'B1' :('BIKE_CLASS', {0:0,1:1,2:0,3:0} , "int" ),
'B2' :('BIKE_CLASS', {0:0,1:0,2:1,3:0} , "int" ),
'B3' :('BIKE_CLASS', {0:0,1:0,2:0,3:1} , "int" ),
'BNE1' :('BIKE_CLASS', {0:1,1:0,2:1,3:1} , "int" ),
'BNE2' :('BIKE_CLASS', {0:1,1:1,2:0,3:1} , "int" ),
'BNE3' :('BIKE_CLASS', {0:1,1:1,2:1,3:0} , "int" ),
'TPER_RISE' :('PER_RISE', ('max',0) , "float" )
}
self['relevant_variables']=['DISTANCE','FT','MTYPE_NUM','TPER_RISE','WRONG_WAY','B0','B1','B2','B3','BNE1','BNE2','BNE3']
for key in changes:
self[key]=changes[key]
class OuterNetworkConfig(UserDict):
"""store network configuration data"""
def __init__(self, changes={}):
UserDict.__init__(self)
self['data_dir']=r"bike_model_input"
self['link_file']=os.path.join(self['data_dir'],'links.csv')
self['node_file']=os.path.join(self['data_dir'],'nodes.csv')
self['dist_var']='DISTANCE'
self['dist_scl']=1/5280 #rescales with node distance x dist_scl= link distance
self['max_centroid']=2454
self['use_dual']=False
self['perform_transformation']=True
self['create_ww_links']=False
self['variable_transforms']={ 'MTYPE_NUM' :('MTYPE', {'SF':1,'MTC':0} , "" )
}
self['relevant_variables']=['DISTANCE','FT','MTYPE_NUM']
for key in changes:
self[key]=changes[key]
class OutputConfig(UserDict):
"""store output configuration data"""
def __init__(self,changes={}):
UserDict.__init__(self)
self['output_dir']=r"bike_model_output"
self['filename_dict']={'pathID':'PathID.csv',
'pathLink':'PathLink.csv',
'estimation_data':'EstimationData.csv',
'bound_data':'Bound.csv',
'overlap_data':'Overlap.csv',
'matrix_data':'Matrix.csv',
'assign_data':'Assign.csv',
'holdback_file':'Holdback.csv'}
self['output_type']=['estimation','bound','geographic','overlap','assign','holdback_sample']
#estimation variable configuration
self['variables']=['DISTANCE',
'B1',
'B2',
'B3',
'TPER_RISE',
'CRIME',
'SPEED',
'WRONG_WAY',
'V_TOT',
'TURN',
'WATERFRONT',
'LANE_OP']
self['aliases']=['DISTANCE',
'BIKE_PCT_1',
'BIKE_PCT_2',
'BIKE_PCT_3',
'AVG_RISE',
'AVG_CRIME',
'AVG_SPEED',
'WRONG_WAY',
'AVG_VOL',
'TURNS',
'WF_PCT',
'AVG_LANES']
self['weights']=[None,
'DISTANCE',
'DISTANCE',
'DISTANCE',
'DISTANCE',
'DISTANCE',
'DISTANCE',
'DISTANCE',
'DISTANCE',
None,
'DISTANCE',
'DISTANCE']
self['trace_funs']=['sum',
'avg',
'avg',
'avg',
'avg',
'avg',
'avg',
'avg',
'avg',
'sum',
'avg',
'avg']
self['final_funs']=[None,None,None,None,None,None,None,None,None,None,None,None]
self['path_size']=True
for key in changes:
self[key]=changes[key]
self.set_time_dir()
def set_time_dir(self):
start_time=time.localtime()
self['time']=''
for i in range(6):
self['time']=self['time']+string.zfill(start_time[i],2)
if i<5:
self['time']=self['time']+'_'
os.mkdir(os.path.join(self['output_dir'],self['time']))
def setup_files(self):
for key in self['filename_dict']:
self[key]=os.path.join(self['output_dir'],self['time'],self['filename_dict'][key])
class ChoiceSetConfig(UserDict):
"""store choice set generation configuration data"""
def __init__(self,changes={},method='doubly_stochastic'):
UserDict.__init__(self)
if method=='link_elimination':
self['method']='link_elimination'
self['master_size']=96
self['consider_size']=96
self['overlap_var']='DISTANCE'
self['only_bound']=False
self['inverted']=False
self['allow_duplicates_of_chosen_route']=True
else:
self['method']='doubly_stochastic'
"""filtering parameters"""
self['overlap_threshold']=0.9 # filter out routes that have an overlap above this
self['overlap_var']='DISTANCE' # network variable for calculating overlap
"""prior coefficient distribution parameters"""
self['ext_bound']=True # use same distribution for each observation (False is deprecated)
self['only_bound']=False # in prepare_estimation.py, only extract prior distribution rather
# than continuing to generate choice sets?
self['bound_from_file']=True # use distribution from file rather than extracting?
self['bound_file']=r'X:\Projects\BikeModel\data\bike_model\input\bound\BoundPredict.csv' #file to use
self['bound_file_override']=True # override variable configuration in this choice_set_config if using file?
self['bounding_box_sample_size']=500 # max number of observations to sample
self['tolerance']=0.01 # percentage threshold to stop binary search when extracting prior distribution
self['log_prior']=True # use log-uniform prior? (False means uniform prior)
"""variable configuration"""
self['variables']=['DISTANCE','BNE1','BNE2','BNE3','WRONG_WAY','TPER_RISE','TURN']
# network variables to use in choice set generation
self['ref']='DISTANCE' # reference variable (coef fixed to 1)
self['ranges']={'BNE1':[0.0000001,1000.0],
'BNE2':[0.0000001,1000.0],
'BNE3':[0.0000001,1000.0],
'TPER_RISE':[0.00001,100000.0],
'WRONG_WAY':[0.0000001,1000.0],
'TURN':[0.0000001,1000.0]} # large initial boundary intervals
self['weights']={'BNE1':'DISTANCE',
'BNE2':'DISTANCE',
'BNE3':'DISTANCE',
'TPER_RISE':'DISTANCE',
'WRONG_WAY':'DISTANCE'} # to multiply each link attribute value by
self['median_compare']=['TURN'] #extract these coefficients with others at their medians (must appear last in self['variables'])
self['randomize_compare']=[] #extract these coefficients with link randomization (must appear last in self['variables'])
"""generate noisy output?"""
self['verbose']=True
"""speed up by randomizing whole network in outer loop and performing searches in inner loop using only comparisons/additions"""
self['inverted']=True # should we?
self['inverted_N_attr']=multiprocessing.cpu_count() # when link attributes were randomized
# individually, this controlled the number of link
# randomizations, now just set it to the number of processors
self['inverted_N_param']=min(2,int(20.0/self['inverted_N_attr']))
# when link attributes were randomized individually,
# this controlled the number of parameters to draw per link
# randomization, now just set it to the number of parameters
# desired divided by the number of processors (e.g. w/;
# N_attr=4 processors x N_param=5 == 20 random parameters)
self['inverted_nested']=False # when link attributes were randomized individually, True
# would nest the attribute and parameter randomization
# loops, now just leave set to False
self['inverted_multiple']=2 # use x times as many random seeds as needed for each source
"""link randomization parameters"""
self['randomize_after']=True # apply link randomization after generalized cost is
# calculated rather than to attributes individually?
# Leave set to True.
self['randomize_after_dev']=0.4 # link randomization scale parameter
self['randomize_after_iters']=3 # number of link randomizations per coefficient
# (e.g. 20 random parameters x 3 randomize_after_iters =
# 60 max choice set size)
"""refrain from filtering out routes that overlap too much with chosen route (used to analyze choice set quality)"""
self['allow_duplicates_of_chosen_route']=False
"""deprecated"""
#parameters used to randomize link attributes individually
self['randomizer_fun']=BetaUnifRandomizer
beta_scl=0.2
unif_dev=0.4
self['randomizer_args']=(2,unif_dev,beta_scl)
self['no_randomize']=['WRONG_WAY','TURN']
#number of generalized cost coefficients to draw if not using inverted loops
self['ds_num_draws']=32
#link randomizer optimization parameters
self['optim_sample_size']=200
self['optim_kappa_vals']=[0.2,0.25]
self['optim_sigma_vals']=[0.4,0.5]
for key in changes:
self[key]=changes[key]
def get_link_randomizer(self,G,master_config):
"""deprecated"""
true_variable_list=list(self['variables'])
if master_config['time_dependent_relation'] is not None:
for var in master_config['time_dependent_relation']:
if var in true_variable_list:
for rule in master_config['time_dependent_relation'][var]:
true_variable_list.append(rule[1])
true_variable_list.remove(var)
link_randomizer=self['randomizer_fun'](G,true_variable_list,self['no_randomize'],*self['randomizer_args'])
return link_randomizer
class AssignConfig(UserDict):
"""compile configuration data"""
def __init__(self, changes={}):
UserDict.__init__(self)
"""how to project outer trips to the county line"""
self['max_inner']=981 # maximum zone id for SF county
self['skip_zones']=[305,313,384,385] # skip these; they don't connect. TODO: make this automatic
self['outer_importance_conditions']=[(982,1348),(2403,2455)] #zones with non-negligible trips to SF
self['boundary_condition']='MTYPE_NUM'# network variable that indicates links which are inside SF
self['outer_impedance']="DISTANCE" # network variable to minimize when projecting trips to county line
"""trip matrices to assign"""
self['matrix_filenames']=[r"bike_model_input\triptable_AM.csv",
#r"bike_model_input\triptable_md.csv",
r"bike_model_input\triptable_PM.csv",
#r"bike_model_input\triptable_ev.csv",
#r"bike_model_input\triptable_ea.csv"
]
self['load_names']=['BIKE_AM','BIKE_PM']#'BIKE_MD','BIKE_EV','BIKE_EA'
"""override bound_file from choice_set_config"""
self['bound_file']=r'bike_BoundPredict.csv'
"""path storage"""
self['pickle_path']='C:/bike_pickle_path' #directory to store path files
self['delete_paths']=False #delete the paths after assignment is complete?
self['load_paths_from_files']=True #use already generated paths rather than starting anew?
"""how to trace variables for utility function"""
self['variables']=['DISTANCE',
'B1',
'B2',
'B3',
'TPER_RISE',
'WRONG_WAY',
'TURN']
self['aliases']=['DISTANCE',
'BIKE_PCT_1',
'BIKE_PCT_2',
'BIKE_PCT_3',
'AVG_RISE',
'WRONG_WAY',
'TURNS_P_MI']
self['weights']=[None,
'DISTANCE',
'DISTANCE',
'DISTANCE',
'DISTANCE',
'DISTANCE',
None]
self['trace_funs']=['sum',
'avg',
'avg',
'avg',
'avg',
'avg',
'sum']
self['final_funs']=[None,None,None,None,None,None,None]
self['path_size']=True
self['path_size_log']=True
self['path_size_alias']='lnpathsize'
self['divisors']={'TURNS_P_MI':'DISTANCE'} # calculate this alias by dividing by this variable
"""fixed coefficients"""
self['fixed_coefficients']=['DISTANCE','TURNS_P_MI','WRONG_WAY','BIKE_PCT_1','BIKE_PCT_2','BIKE_PCT_3','AVG_RISE','lnpathsize']
self['alpha']=[-1.05,-0.21,-13.30,1.89,2.15,0.35,-154.0,1.0]
"""random coefficients"""
self['use_random_coefficients']=False
self['random_coefficients']=[]#['BIKE_PCT_1','BIKE_PCT_2','BIKE_PCT_3','AVG_RISE']
self['random_transformations']=[]#[idenfun,idenfun,idenfun,idenfun]
self['latent_mu']=[]#[1.82,2.49,0.76,-2.22]
self['latent_sigma']=array([])
"""array( [ [24.95, 0., 6.58, 0. ],
[0., 5.45, 2.91, 0. ],
[0., 0., 4.19, 0. ],
[0., 0., 0., 3.85 ] ] )"""
self['mixing_granularity']=0.2 # number of trips to simulate as an individual
"""for debugging code"""
self['test_small_matrix']=False
for key in changes:
self[key]=changes[key]
def idenfun(x):
return x
def negexp(x):
return -exp(x)