def eval_one(step):
eval_features = []
for i in range(0, len(all_features)):
if step[i]:
eval_features.append(all_features[i])
all_observations = []
all_pred_combined = []
Y = []
P = []
for group in range(0, len(groups)):
train_stations, test_stations = generate_train_test_station_list(
group, groups)
train_station_set = set([float(s) for s in train_stations])
test_station_set = set([float(s) for s in test_stations])
train_lower = [
float(train_stations[i]) for i in range(0, len(train_stations))
if i < (len(train_stations) / 2.0)
]
train_lower_set = set(train_lower)
train_upper = [
float(train_stations[i]) for i in range(0, len(train_stations))
if i >= (len(train_stations) / 2.0)
]
train_upper_set = set(train_upper)
test_lower = [
float(test_stations[i]) for i in range(0, len(test_stations))
if i < (len(test_stations) / 2.0)
]
# tw_lower
trainX, testX, trainY, testY = splitDataForXValidation(
train_lower_set, test_station_set, "location", data, tw_features,
"target")
model = create_model()
model.fit(trainX, trainY)
prediction_lower = model.predict(testX)
# tw_upper
trainX, testX, trainY, testY = splitDataForXValidation(
train_upper_set, test_station_set, "location", data, tw_features,
"target")
model = create_model()
model.fit(trainX, trainY)
prediction_upper = model.predict(testX)
trainX, testX, trainY, testY, train_location, test_location = splitDataForXValidationWithLocation(
train_station_set, test_station_set, "location", data,
eval_features, "target")
train_label = generate_label(train_location, train_lower)
test_label = generate_label(test_location, test_lower)
model = create_classifier_model()
model.fit(trainX, train_label)
prediction_label = model.predict(testX)
pred_combined = generate_combined_prediction(prediction_label,
prediction_lower,
prediction_upper)
all_pred_combined.extend(pred_combined)
all_observations.extend(testY)
Y.extend(test_label)
P.extend(prediction_label)
rmse = rmseEval(all_observations, all_pred_combined)[1]
accuracy = accuracy_score(Y, P)
return rmse, accuracy
def eval_one(step):
if step in cached_results:
return cached_results[step]
eval_features = []
for i in range(0, len(all_features)):
if step[i]:
eval_features.append(all_features[i])
all_observations = []
all_pred_combined = []
for group in range(0, len(groups)):
train_stations, test_stations = generate_train_test_station_list(
group, groups)
train_station_set = set([float(s) for s in train_stations])
test_station_set = set([float(s) for s in test_stations])
train_lower = [
float(train_stations[i]) for i in range(0, len(train_stations))
if i < (len(train_stations) / 2.0)
]
train_lower_set = set(train_lower)
train_upper = [
float(train_stations[i]) for i in range(0, len(train_stations))
if i >= (len(train_stations) / 2.0)
]
train_upper_set = set(train_upper)
test_lower = [
float(test_stations[i]) for i in range(0, len(test_stations))
if i < (len(test_stations) / 2.0)
]
# tw_lower
trainX, testX, trainY, testY = splitDataForXValidation(
train_lower_set, test_station_set, "location", data, tw_features,
"target")
model = create_model()
model.fit(trainX, trainY)
prediction_lower = model.predict(testX)
# tw_upper
trainX, testX, trainY, testY = splitDataForXValidation(
train_upper_set, test_station_set, "location", data, tw_features,
"target")
model = create_model()
model.fit(trainX, trainY)
prediction_upper = model.predict(testX)
trainX, testX, trainY, testY, train_location, test_location = splitDataForXValidationWithLocation(
train_station_set, test_station_set, "location", data,
eval_features, "target")
train_label = generate_label(train_location, train_lower)
test_label = generate_label(test_location, test_lower)
model = create_classifier_model()
model.fit(trainX, train_label)
prediction_label = model.predict(testX)
pred_combined = generate_combined_prediction(prediction_label,
prediction_lower,
prediction_upper)
all_pred_combined.extend(pred_combined)
all_observations.extend(testY)
rmse = rmseEval(all_observations, all_pred_combined)[1]
cached_results[step] = rmse
# save down the cached result
cache_output = open(CACHE_FILE, "a")
step_list = [str(s) for s in step]
step_str = ",".join(step_list)
cache_output.write(str(rmse) + ";" + step_str + "\n")
cache_output.close()
return rmse
all_pred_TWA = []
all_pred_combined = []
all_test_location = []
for group in range(0, len(groups)):
log("group: " + str(group + 1))
train_stations, test_stations = generate_train_test_station_list(
group, groups)
log("\ttrain_stations: " + str(train_stations))
log("\ttest_stations: " + str(test_stations))
train_station_set = set([float(s) for s in train_stations])
test_station_set = set([float(s) for s in test_stations])
trainX, testX, trainY, testY, trainLocation, testLocation = splitDataForXValidationWithLocation(
train_station_set, test_station_set, "location", data, all_features,
"target")
model = RandomForestRegressor(min_samples_leaf=29,
n_estimators=64,
n_jobs=-1,
random_state=42)
model.fit(trainX, trainY)
prediction_TW = model.predict(testX)
rmse = rmseEval(testY, prediction_TW)[1]
log("\tALL rmse: " + str(rmse))
all_observations.extend(testY)
all_pred_ALL.extend(prediction_TW)
all_test_location.extend(testLocation)
trainX, testX, trainY, testY, trainLocation, testLocation = splitDataForXValidationWithLocation(
train_station_set, test_station_set, "location", data, tw_features,