def grade_session_with_automatic_model(request,session_code):
session = get_object_or_404(Session,code=session_code)
task_name = session.task_def.name;
ag_model_fn = os.path.join(AUTOGRADING_MODEL_ROOT, task_name, 'ag_grading.bin');
if not os.path.exists(ag_model_fn):
return HttpResponse("No model."+ag_model_fn)
temp_root=os.path.join('/var/tmp/autograding',session_code)
if not os.path.exists(temp_root):
os.makedirs(temp_root)
create_flat_file.create_grading_flat_file(session.code,temp_root, [0,0,1])
build_te_model.make_predictions_with_model(temp_root,ag_model_fn)
(predictions,gt_url) = round_predictions.round_predictions(temp_root)
(wrk,is_created)=Worker.objects.get_or_create(worker="AUTO-ADDITIVE-GROVES");
resp=HttpResponse("Done.")
for pred, ref in map(None,predictions,gt_url):
submission_id = int(ref.split('\t')[0])
resp.write("%d\t%d\n" % (submission_id,pred))
subm = get_object_or_404(SubmittedTask,id=submission_id)
(rcd,created)=ManualGradeRecord.objects.get_or_create(submission=subm,
worker=wrk);
rcd.feedback="Computer-generated grade"
rcd.quality=pred;
rcd.save();
return resp
def grade_session_with_automatic_model(request, session_code):
session = get_object_or_404(Session, code=session_code)
task_name = session.task_def.name
ag_model_fn = os.path.join(AUTOGRADING_MODEL_ROOT, task_name, "ag_grading.bin")
if not os.path.exists(ag_model_fn):
return HttpResponse("No model." + ag_model_fn)
temp_root = os.path.join("/var/tmp/autograding", session_code)
if not os.path.exists(temp_root):
os.makedirs(temp_root)
create_flat_file.create_grading_flat_file(session.code, temp_root, [0, 0, 1])
build_te_model.make_predictions_with_model(temp_root, ag_model_fn)
(predictions, gt_url) = round_predictions.round_predictions(temp_root)
(wrk, is_created) = Worker.objects.get_or_create(worker="AUTO-ADDITIVE-GROVES")
resp = HttpResponse("Done.")
for pred, ref in map(None, predictions, gt_url):
submission_id = int(ref.split("\t")[0])
resp.write("%d\t%d\n" % (submission_id, pred))
subm = get_object_or_404(SubmittedTask, id=submission_id)
(rcd, created) = ManualGradeRecord.objects.get_or_create(submission=subm, worker=wrk)
rcd.feedback = "Computer-generated grade"
rcd.quality = pred
rcd.save()
return resp
def main(argv):
try:
opts, args = getopt.getopt(argv, "ho:", ["output="])
except getopt.GetoptError:
print 'xgb.py [-o [2008] [2012]]'
sys.exit(2)
for opt, arg in opts:
if opt == '-h':
print 'xgb.py [-o [2008] [2012]]'
sys.exit()
elif opt in ("-o", "--output"):
if (arg == "2008"):
preds = round_predictions(xgb_model.predict(X_test_2008))
# DEBUG
print("preds.shape" + str(preds.shape))
make_submission_2008("submissions/xgb_2008.csv", preds)
elif (arg == "2012"):
preds = round_predictions(xgb_model.predict(X_test_2012))
make_submission_2012("submissions/xgb_2012.csv", preds)
def build_model_from_session(request,session_code):
session = get_object_or_404(Session,code=session_code)
task_name = session.task_def.name;
ag_model_fn = os.path.join(AUTOGRADING_MODEL_ROOT, task_name, 'ag_grading.bin');
ag_training_root = os.path.join(AUTOGRADING_MODEL_ROOT, task_name, 'training');
if not os.path.exists(ag_training_root):
#os.rmdir(ag_training_root);
os.makedirs(ag_training_root);
temp_root=os.path.join('/var/tmp/autograding',session_code)
if not os.path.exists(temp_root):
os.makedirs(temp_root)
resp=HttpResponse("Building grading model:\n",mimetype="text/plain")
resp.write("Creating data file\n")
resp.flush()
create_flat_file.create_grading_flat_file(session.code, ag_training_root, [0.5,0.25,0.25])
resp.write(" done\n")
resp.write("Learning the model on train+val\n")
resp.flush()
build_te_model.build_model(ag_training_root);
resp.write(" done\n")
ag_trained_model_fn=os.path.join(ag_training_root,'model.bin')
build_te_model.make_predictions_with_model(ag_training_root,ag_trained_model_fn)
resp.write("Predicting on the testset\n")
resp.flush()
round_predictions.round_predictions(ag_training_root)
resp.write(" done\n")
resp.write("----------------\n")
resp.write("Testset report\n")
rpt_fn=os.path.join(ag_training_root,'preds.txt.report.txt')
for s in open(rpt_fn,'r').readlines():
resp.write(s)
shutil.copyfile(ag_trained_model_fn,ag_model_fn)
return resp
def build_model_from_session(request, session_code):
session = get_object_or_404(Session, code=session_code)
task_name = session.task_def.name
ag_model_fn = os.path.join(AUTOGRADING_MODEL_ROOT, task_name, "ag_grading.bin")
ag_training_root = os.path.join(AUTOGRADING_MODEL_ROOT, task_name, "training")
if not os.path.exists(ag_training_root):
# os.rmdir(ag_training_root);
os.makedirs(ag_training_root)
temp_root = os.path.join("/var/tmp/autograding", session_code)
if not os.path.exists(temp_root):
os.makedirs(temp_root)
resp = HttpResponse("Building grading model:\n", mimetype="text/plain")
resp.write("Creating data file\n")
resp.flush()
create_flat_file.create_grading_flat_file(session.code, ag_training_root, [0.5, 0.25, 0.25])
resp.write(" done\n")
resp.write("Learning the model on train+val\n")
resp.flush()
build_te_model.build_model(ag_training_root)
resp.write(" done\n")
ag_trained_model_fn = os.path.join(ag_training_root, "model.bin")
build_te_model.make_predictions_with_model(ag_training_root, ag_trained_model_fn)
resp.write("Predicting on the testset\n")
resp.flush()
round_predictions.round_predictions(ag_training_root)
resp.write(" done\n")
resp.write("----------------\n")
resp.write("Testset report\n")
rpt_fn = os.path.join(ag_training_root, "preds.txt.report.txt")
for s in open(rpt_fn, "r").readlines():
resp.write(s)
shutil.copyfile(ag_trained_model_fn, ag_model_fn)
return resp
def pred_2008():
lasso_unrounded = modified_predict(lasso, X_test_2008)
ridge_unrounded = modified_predict(ridge, X_test_2008)
mlp_unrounded = modified_predict(mlp, X_test_2008)
xgb_unrouned = xgb_2008.reshape(-1, 1)
rand_forest_unrounded = modified_predict(rand_forest, X_test_2008)
adaboost_ran_forest_unrounded = modified_predict(adaboost_ran_forest,
X_test_2008)
adaboost_unrounded = modified_predict(adaboost, X_test_2008)
total = xgb_score + adaboost_ran_forest_score + adaboost_score + mlp_score + \
lasso_score + rand_forest_score + ridge_score
lasso_weight = lasso_score / total
ridge_weight = ridge_score / total
xgb_weight = xgb_score / total
adaboost_ran_forest_weight = adaboost_ran_forest_score / total
adaboost_weight = adaboost_score / total
mlp_weight = mlp_score / total
rand_forest_weight = rand_forest_score / total
lasso_2008 = lasso_weight * lasso_unrounded
ridge_2008 = ridge_weight * ridge_unrounded
xgb_2008_weighted = xgb_weight * xgb_unrouned
adaboost_ran_forest_2008 = adaboost_ran_forest_weight * adaboost_ran_forest_unrounded
adaboost_2008 = adaboost_weight * adaboost_unrounded
mlp_2008 = mlp_weight * mlp_unrounded
rand_forest_2008 = rand_forest_weight * rand_forest_unrounded
lasso_unrounded = 0
ridge_unrounded = 0
mlp_unrounded = 0
xgb_unrouned = 0
rand_forest_unrounded = 0
adaboost_ran_forest_unrounded = 0
adaboost_unrounded = 0
print "Starting Adding"
temp1 = np.add(np.add(lasso_2008, ridge_2008), xgb_2008_weighted)
print "Halway Through Adding"
temp2 = np.add(
np.add(np.add(adaboost_ran_forest_2008, adaboost_2008), mlp_2008),
rand_forest_2008)
ensemble = np.add(temp1, temp2)
print "Done Adding!"
temp1 = 0
temp2 = 0
print ensemble
ensemble = round_predictions(ensemble)
print "Min of ensemble: ", np.min(ensemble), ". Max: ", np.max(ensemble)
return ensemble
def main(argv):
try:
opts, args = getopt.getopt(argv, "ho:", ["output="])
except getopt.GetoptError:
print 'xgb.py [-o [2008] [2012]]'
sys.exit(2)
for opt, arg in opts:
if opt == '-h':
print 'xgb.py [-o [2008] [2012]]'
sys.exit()
elif opt in ("-o", "--output"):
if (arg == "2008"):
preds = round_predictions(xgb_model.predict(X_test_2008))
# DEBUG
print("preds.shape" + str(preds.shape))
make_submission_2008("submissions/xgb_2008.csv", preds)
elif (arg == "2012"):
preds = round_predictions(xgb_model.predict(X_test_2012))
make_submission_2012("submissions/xgb_2012.csv", preds)
if __name__ == "__main__":
main(sys.argv[1:])
preds = round_predictions(xgb_model.predict(X_train_2008))
train_error = np.mean(preds == Y_train_2008)
print("XGB train error = " + str(train_error))
ver_preds = round_predictions(xgb_model.predict(X_ver_2008))
ver_error = np.mean(ver_preds == Y_ver_2008)
print("XGB ver error = " + str(ver_error))
lasso = read_make_pkl("saved_objs/lasso.pkl", gen_lasso)
def main(argv):
try:
opts, args = getopt.getopt(argv, "ho:", ["output="])
except getopt.GetoptError:
print 'lasso.py [-o [2008] [2012]]'
sys.exit(2)
for opt, arg in opts:
if opt == '-h':
print 'lasso.py [-o [2008] [2012]]'
sys.exit()
elif opt in ("-o", "--output"):
if (arg == "2008"):
preds = round_predictions(lasso.predict(X_test_2008))
# DEBUG
print("lasso.predict(X_test_2008).shape" +
str(lasso.predict(X_test_2008).shape))
make_submission_2008("submissions/lasso_2008.csv", preds)
elif (arg == "2012"):
preds = round_predictions(lasso.predict(X_test_2012))
make_submission_2012("submissions/lasso_2012.csv", preds)
if __name__ == "__main__":
main(sys.argv[1:])
preds = round_predictions(lasso.predict(X_train_2008))
train_error = np.mean(preds == Y_train_2008)
print("Lasso train error = " + str(train_error))
def lin_reg_modified_predict(model, X):
preds = model.predict(X).reshape(-1, 1)
mpreds = round_predictions(preds)
# Debug
print("mpreds.shape: " + str(mpreds.shape))
return mpreds
ridge = read_make_pkl("saved_objs/ridge.pkl", gen_ridge)
def main(argv):
try:
opts, args = getopt.getopt(argv, "ho:", ["output="])
except getopt.GetoptError:
print 'ridge.py [-o [2008] [2012]]'
sys.exit(2)
for opt, arg in opts:
if opt == '-h':
print 'ridge.py [-o [2008] [2012]]'
sys.exit()
elif opt in ("-o", "--output"):
if (arg == "2008"):
preds = round_predictions(ridge.predict(X_test_2008))
# DEBUG
print("ridge.predict(X_test_2008).shape" +
str(ridge.predict(X_test_2008).shape))
make_submission_2008("submissions/ridge_2008.csv", preds)
elif (arg == "2012"):
preds = round_predictions(ridge.predict(X_test_2012))
make_submission_2012("submissions/ridge_2012.csv", preds)
if __name__ == "__main__":
main(sys.argv[1:])
preds = round_predictions(ridge.predict(X_train_2008))
train_error = np.mean(preds == Y_train_2008)
print("Ridge train error = " + str(train_error))