import sys
from seo import letor_fold_creator_min_max_normalize as lfc
from model_running import cross_validator as cv
from seo import exponential_budget_cost_creator as e
if __name__ == "__main__":
data_set_location = "/lv_local/home/sgregory/letor"
new_data_set_location = "/lv_local/home/sgregory/letor_fixed1"
l = lfc.letor_folds_creator(data_set_location, new_data_set_location, True)
c = cv.cross_validator(5, l, "LTOR1")
c.k_fold_cross_validation("SVM", "qrels")
from seo import query_to_fold as qtf
from seo import letor_fold_creator as lfc
from model_running import cross_validator as cv
import sys
if __name__ == "__main__":
data_set_location = sys.argv[1]
print data_set_location
new_data_set_location = sys.argv[2]
qrel_path = sys.argv[3]
print qrel_path
q = qtf.qtf(data_set_location)
q.create_query_to_fold_index()
l = lfc.letor_folds_creator(data_set_location, new_data_set_location, True)
c = cv.cross_validator(5, l, "LTOR_MART")
c.k_fold_cross_validation("LAMBDAMART", qrel_path)
def average_list(list_a,iterations):
return [float(a)/iterations for a in list_a]
def sum_dicts(x,y):
return {k: x.get(k, 0) + y.get(k, 0) for k in set(x) | set(y)}
def average_dict(dict_a,iterations):
return {k: float(dict_a.get(k, 0))/iterations for k in set(dict_a)}
if __name__ == "__main__":
data_set_location = "/lv_local/home/sgregory/letor_fixed1"
q = qtf.qtf(data_set_location)
q.create_query_to_fold_index()
l = lfc.letor_folds_creator_z_normalize(data_set_location, data_set_location, True)
c = cv.cross_validator(5, l, "LTOR_MART_min_max")
lbda_score_file = "/lv_local/home/sgregory/LTOR_MART_min_max/test_scores_trec_format/LAMBDAMART/final_score_combined.txt"
svm_score_file = "/lv_local/home/sgregory/LTOR1/test_scores_trec_format/SVM/final_score_combined.txt"
rel_index = ri.relevance_index("qrels")
rel_index.create_relevance_index()
pool = p(3)
gg = d.lambda_mart_stats_handler("01", 0.1,c)
aa = d.lambda_mart_stats_handler("005", 0.05,c)
bb = d.lambda_mart_stats_handler("001", 0.01, c)
svm_gg = srfh.winner_reference_point_random("01",0.1)
svm_aa = srfh.winner_reference_point_random("005", 0.05)
svm_bb = srfh.winner_reference_point_random("001", 0.01)
lbda_chosen_models = gg.recover_models_per_fold("/lv_local/home/sgregory/LTOR_MART_min_max/models/LAMBDAMART",
"/lv_local/home/sgregory/LTOR_MART_min_max/test_scores_trec_format/LAMBDAMART/")
import os
import sys
from model_running import cross_validator as cv
if __name__ == "__main__":
model = sys.argv[1] #user's input of model
if model != "LAMBDAMART" and model != "SVM":
print("please insert correct model to run")
sys.exit(1)
featues_file = sys.argv[2]
if not os.path.exists(featues_file):
print("please insert correct path to train file")
sys.exit(1)
query_relevance_file = sys.argv[3]
if not os.path.exists(query_relevance_file):
print("please insert correct path to relevance file")
sys.exit(1)
data_set = sys.argv[4]
cross_validator = cv.cross_validator(
5, featues_file, data_set,
200) #TODO: maybe add user input params for more generality
cross_validator.k_fold_cross_validation(model, query_relevance_file)