def create_predict_table(LCM, agent_set, agents_index, observed_choices_id, data_objects, geographies=[]):
resources = data_objects
mc_choices = sample_choice(LCM.model.probabilities) # monte carlo choice
mc_choices_index = LCM.model_resources.translate("index")[mc_choices]
maxprob_choices = sample_choice(LCM.model.probabilities, method="max_prob") # max prob choice
maxprob_choices_index = LCM.model_resources.translate("index")[maxprob_choices]
results = []
gcs = resources.translate("gridcell")
for geography in geographies:
geos = resources.translate(geography)
# get observed geo_id
obs = copy_dataset(agent_set)
obs.subset_by_index(agents_index)
obs.set_values_of_one_attribute(gcs.id_name[0], observed_choices_id)
resources.merge({"household": obs}) # , "gridcell": gcs, "zone": zones, "faz":fazes})
obs.compute_variables(geos.id_name[0], resources=resources)
obs_geo_ids = obs.get_attribute(geos.id_name[0])
# count simulated choices
sim = copy_dataset(obs)
sim.set_values_of_one_attribute(gcs.id_name[0], gcs.get_id_attribute()[mc_choices_index])
resources.merge({"household": sim})
geos_size = geos.size()
geo_ids = geos.get_id_attribute()
pred_matrix = zeros((geos_size, geos_size))
p_success = zeros((geos_size,)).astype(Float32)
f = 0
for geo_id in geo_ids:
index_in_geo = where(obs_geo_ids == geo_id)[0]
resources.merge({"select_index": index_in_geo})
geos.compute_variables("number_of_select_households", resources=resources)
pred_matrix[f] = geos.get_attribute("number_of_select_households")
if sum(pred_matrix[f]) > 0:
p_success[f] = float(pred_matrix[f, f]) / sum(pred_matrix[f])
sim.increment_version("grid_id") # to trigger recomputation in next iteration
f += 1
print p_success
results.append((pred_matrix.copy(), p_success.copy()))
return results
def create_predict_table(LCM, agent_set, agents_index, observed_choices_id, data_objects, geographies=[]):
resources = data_objects
mc_choices = sample_choice(LCM.model.probabilities) #monte carlo choice
mc_choices_index = LCM.model_resources.translate("index")[mc_choices]
maxprob_choices = sample_choice(LCM.model.probabilities, method="max_prob") #max prob choice
maxprob_choices_index = LCM.model_resources.translate("index")[maxprob_choices]
results = []
gcs = resources.translate("gridcell")
for geography in geographies:
geos = resources.translate(geography)
#get observed geo_id
obs = copy_dataset(agent_set)
obs.subset_by_index(agents_index)
obs.set_values_of_one_attribute(gcs.id_name[0], observed_choices_id)
resources.merge({"household": obs}) #, "gridcell": gcs, "zone": zones, "faz":fazes})
obs.compute_variables(geos.id_name[0], resources=resources)
obs_geo_ids = obs.get_attribute(geos.id_name[0])
#count simulated choices
sim = copy_dataset(obs)
sim.set_values_of_one_attribute(gcs.id_name[0], gcs.get_id_attribute()[mc_choices_index])
resources.merge({"household": sim})
geos_size = geos.size()
geo_ids = geos.get_id_attribute()
pred_matrix = zeros((geos_size, geos_size))
p_success = zeros((geos_size,)).astype(Float32)
f = 0
for geo_id in geo_ids:
index_in_geo = where(obs_geo_ids == geo_id)[0]
resources.merge({"select_index": index_in_geo})
geos.compute_variables("number_of_select_households", resources=resources)
pred_matrix[f] = geos.get_attribute("number_of_select_households")
if sum(pred_matrix[f]) > 0:
p_success[f] = float(pred_matrix[f, f])/sum(pred_matrix[f])
sim.increment_version('grid_id') #to trigger recomputation in next iteration
f += 1
print p_success
results.append((pred_matrix.copy(), p_success.copy()))
return results
def create_prediction_success_table(
LCM, location_set, observed_choices_id, geographies=[], choice_method="mc", data_objects=None
):
"""this function creates a table tabulating number of agents observed versus predicted by geographies for location choice model
LCM is an instance of Location Choice Model after run_estimation,
location_set is the set of location in simulation, e.g. gridcell,
observed_choice_id is the location_set id (e.g. grid_id) observed,
geographies is a list of geographies to create prediction sucess table for,
choice_method is the method used to select choice for agents, either mc or max_prob
data_objects is the same as data_objects used to run LCM simulation, but includes entries for geographies
"""
LCM.simulate_step()
choices = sample_choice(LCM.model.probabilities, choice_method)
choices_index = LCM.model_resources.translate("index")[choices] # translate choices into index of location_set
# maxprob_choices = sample_choice(LCM.model.probabilities, method="max_prob") #max prob choice
# maxprob_choices_index = LCM.model_resources.translate("index")[maxprob_choices]
results = []
gcs = location_set
for geography in geographies:
geo = data_objects.translate(geography)
# get geo_id for observed agents
gc_index = gcs.get_id_index(observed_choices_id)
if geo.id_name[0] not in gcs.get_attribute_names():
gcs.compute_variables(geo.id_name[0], resources=data_objects)
geo_ids_obs = gcs.get_attribute(geo.id_name[0])[gc_index]
# obs = copy.deepcopy(agent_set)
# obs.subset_by_index(agents_index)
# obs.set_values_of_one_attribute(gcs.id_name[0], observed_choices_id)
# resources.merge({"household": obs}) #, "gridcell": gcs, "zone": zones, "faz":fazes})
# obs.compute_variables(geo.id_name[0], resources=resources)
# obs_geo_ids = obs.get_attribute(geo.id_name[0])
# get geo_id for simulated agents
geo_ids_sim = gcs.get_attribute(geo.id_name[0])[choices_index]
# sim = copy_dataset(obs)
# sim.set_values_of_one_attribute(gcs.id_name[0], gcs.get_id_attribute()[mc_choices_index])
# resources.merge({"household": sim})
geo_size = geo.size()
myids = geo.get_id_attribute()
pred_matrix = zeros((geo_size, geo_size))
p_success = zeros((geo_size,)).astype(Float32)
f = 0
for geo_id in myids:
ids = geo_ids_sim[where(geo_ids_obs == geo_id)] # get simulated geo_id for agents observed in this geo_id
# resources.merge({"agents_index": agents_index_in_geo, "agent":sim})
what = ones(ids.size())
pred_matrix[f] = array(nd_image_sum(what, labels=ids, index=myids))
print pred_matrix[f]
if sum(pred_matrix[f]) > 0:
p_success[f] = float(pred_matrix[f, f]) / sum(pred_matrix[f])
# sim.increment_version(gcs.id_name[0]) #to trigger recomputation in next iteration
f += 1
print p_success
results.append((pred_matrix.copy(), p_success.copy()))
return results
def create_prediction_success_table(LCM, location_set, observed_choices_id, geographies=[], \
choice_method='mc', data_objects=None):
"""this function creates a table tabulating number of agents observed versus predicted by geographies for location choice model
LCM is an instance of Location Choice Model after run_estimation,
location_set is the set of location in simulation, e.g. gridcell,
observed_choice_id is the location_set id (e.g. grid_id) observed,
geographies is a list of geographies to create prediction sucess table for,
choice_method is the method used to select choice for agents, either mc or max_prob
data_objects is the same as data_objects used to run LCM simulation, but includes entries for geographies
"""
LCM.simulate_step()
choices = sample_choice(LCM.model.probabilities, choice_method)
choices_index = LCM.model_resources.translate("index")[
choices] #translate choices into index of location_set
#maxprob_choices = sample_choice(LCM.model.probabilities, method="max_prob") #max prob choice
#maxprob_choices_index = LCM.model_resources.translate("index")[maxprob_choices]
results = []
gcs = location_set
for geography in geographies:
geo = data_objects.translate(geography)
#get geo_id for observed agents
gc_index = gcs.get_id_index(observed_choices_id)
if geo.id_name[0] not in gcs.get_attribute_names():
gcs.compute_variables(geo.id_name[0], resources=data_objects)
geo_ids_obs = gcs.get_attribute(geo.id_name[0])[gc_index]
# obs = copy.deepcopy(agent_set)
# obs.subset_by_index(agents_index)
# obs.set_values_of_one_attribute(gcs.id_name[0], observed_choices_id)
#resources.merge({"household": obs}) #, "gridcell": gcs, "zone": zones, "faz":fazes})
# obs.compute_variables(geo.id_name[0], resources=resources)
# obs_geo_ids = obs.get_attribute(geo.id_name[0])
#get geo_id for simulated agents
geo_ids_sim = gcs.get_attribute(geo.id_name[0])[choices_index]
#sim = copy_dataset(obs)
#sim.set_values_of_one_attribute(gcs.id_name[0], gcs.get_id_attribute()[mc_choices_index])
#resources.merge({"household": sim})
geo_size = geo.size()
myids = geo.get_id_attribute()
pred_matrix = zeros((geo_size, geo_size))
p_success = zeros((geo_size, )).astype(Float32)
f = 0
for geo_id in myids:
ids = geo_ids_sim[where(
geo_ids_obs == geo_id
)] #get simulated geo_id for agents observed in this geo_id
#resources.merge({"agents_index": agents_index_in_geo, "agent":sim})
what = ones(ids.size())
pred_matrix[f] = array(nd_image_sum(what, labels=ids, index=myids))
print pred_matrix[f]
if sum(pred_matrix[f]) > 0:
p_success[f] = float(pred_matrix[f, f]) / sum(pred_matrix[f])
#sim.increment_version(gcs.id_name[0]) #to trigger recomputation in next iteration
f += 1
print p_success
results.append((pred_matrix.copy(), p_success.copy()))
return results
