Skip to content

Commit

Permalink
test pull nono
Browse files Browse the repository at this point in the history 
Co-authored-by: JYeonKim <JYeonKim@users.noreply.github.com>
  • Loading branch information
@hyeon9698
hyeon9698 committed Nov 23, 2021
1 parent 8c264e2 commit c44d487
Show file tree
Hide file tree
Showing 2 changed files with 89 additions and 89 deletions.
176 changes: 88 additions & 88 deletions OJ-BE/aicontest/views/oj.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -15,7 +15,7 @@
import json
import csv
# import pandas as pd
import numpy as np
# import numpy as np
# import sklearn

import logging
Expand Down Expand Up @@ -219,98 +219,98 @@ def post(self, request):
2,13
3,346
"""
y_true = np.array(np.loadtxt(os.path.join(settings.SOLUTION_DIR, csv.solution_id, "solution.csv"), delimiter=",", dtype=np.float32, skiprows=1, usecols = (1,)))
y_pred = np.array(np.loadtxt(os.path.join(settings.PREDICT_DIR, predict_id, "predict.csv"), delimiter=",", dtype=np.float32, skiprows=1, usecols = (1,)))
logger.info("y_true={}".format(str(y_true)))
logger.info("y_pred={}".format(str(y_pred)))
# y_true = np.array(np.loadtxt(os.path.join(settings.SOLUTION_DIR, csv.solution_id, "solution.csv"), delimiter=",", dtype=np.float32, skiprows=1, usecols = (1,)))
# y_pred = np.array(np.loadtxt(os.path.join(settings.PREDICT_DIR, predict_id, "predict.csv"), delimiter=",", dtype=np.float32, skiprows=1, usecols = (1,)))
# logger.info("y_true={}".format(str(y_true)))
# logger.info("y_pred={}".format(str(y_pred)))

eval_type = csv.eval_type
logger.info("eval_type={}".format(eval_type))

if eval_type == 'mse':
logger.info("if mse in")
y_score = np.square(np.subtract(y_true, y_pred)).mean()
logger.info("y_score={}".format(str(y_score)))

if eval_type == 'acc':
logger.info("if acc in")
y_score = (y_true == y_pred).mean()
logger.info("y_score={}".format(str(y_score)))

if eval_type == 'rmse':
logger.info("if rmse in")
y_score = np.sqrt(np.mean(np.subtract(y_true, y_pred)**2))
# eval_type = csv.eval_type
# logger.info("eval_type={}".format(eval_type))

# if eval_type == 'mse':
# logger.info("if mse in")
# y_score = np.square(np.subtract(y_true, y_pred)).mean()
# logger.info("y_score={}".format(str(y_score)))

# if eval_type == 'acc':
# logger.info("if acc in")
# y_score = (y_true == y_pred).mean()
# logger.info("y_score={}".format(str(y_score)))

# if eval_type == 'rmse':
# logger.info("if rmse in")
# y_score = np.sqrt(np.mean(np.subtract(y_true, y_pred)**2))

if eval_type == 'auc':
logger.info("if auc in")

if eval_type == 'mean_f_score':
logger.info("if mean_f_score in")
# accuracy = np.mean(np.equal(y_true, y_pred))
right = np.sum(y_true * y_pred == 1)
precision = right / np.sum(y_pred)
recall = right / np.sum(y_true)
f1 = 2 * precision*recall/(precision+recall)
y_score = f1
logger.info("if mean_f_score out")

if eval_type == 'mae':
logger.info("if mae in")
y_score = np.mean(np.abs(np.subtract(y_true, y_pred)))

if eval_type == 'f_beta':
logger.info("if f_beta in")

if eval_type == 'logloss':
logger.info("if logloss in")
def my_logloss(answer_array, proba_array):
# 출처 https://velog.io/@skyepodium/logloss-%EC%95%8C%EC%95%84%EB%B3%B4%EA%B8%B0
# 0이면 무한대 값이 나오기 때문에 0에 가까운 값으로 치환해줍니다.
MIN_VALUE = 1e-15
# array의 크기를 가져옵니다.
size = answer_array.shape[0]
# 반복문을 사용해서 logloss의 합을 계산합니다.
logloss_sum = 0
# zip함수로 묶으면 함께 순회할 수 있습니다.
for answer, arr in zip(answer_array, proba_array):
proba = arr[answer - 1]
# 0이면 무한대 값이 나오기 때문에 0에 가까운 값으로 치환해줍니다.
if proba <= MIN_VALUE:
proba = MIN_VALUE
# 음의 로그함수에 넣어서 logloss 계산
logloss_sum += -np.log(proba)
# logloss의 평균 계산
result = logloss_sum / size
# 반환
return result
y_score = my_logloss(y_true, y_pred)

if eval_type == 'rmsle':
logger.info("if rmsle in")
def rmsle(predicted_values, actual_values):
# 예측값과 실제 값에 1을 더하고 로그를 씌워준다.
log_predict = np.log(predicted_values + 1)
log_actual = np.log(actual_values + 1)
# 위에서 계산한 예측값에서 실제값을 빼주고 제곱을 해준다.
difference = log_predict - log_actual
# difference = (log_predict - log_actual) ** 2
difference = np.square(difference)
# 평균을 낸다.
mean_difference = difference.mean()
# 다시 루트를 씌운다.
score = np.sqrt(mean_difference)
return score
y_score = rmsle(y_true, y_pred)

if eval_type == 'map@k':
logger.info("if map@k in")
# if eval_type == 'auc':
# logger.info("if auc in")

# if eval_type == 'mean_f_score':
# logger.info("if mean_f_score in")
# # accuracy = np.mean(np.equal(y_true, y_pred))
# right = np.sum(y_true * y_pred == 1)
# precision = right / np.sum(y_pred)
# recall = right / np.sum(y_true)
# f1 = 2 * precision*recall/(precision+recall)
# y_score = f1
# logger.info("if mean_f_score out")

# if eval_type == 'mae':
# logger.info("if mae in")
# y_score = np.mean(np.abs(np.subtract(y_true, y_pred)))

# if eval_type == 'f_beta':
# logger.info("if f_beta in")

# if eval_type == 'logloss':
# logger.info("if logloss in")
# def my_logloss(answer_array, proba_array):
# # 출처 https://velog.io/@skyepodium/logloss-%EC%95%8C%EC%95%84%EB%B3%B4%EA%B8%B0
# # 0이면 무한대 값이 나오기 때문에 0에 가까운 값으로 치환해줍니다.
# MIN_VALUE = 1e-15
# # array의 크기를 가져옵니다.
# size = answer_array.shape[0]
# # 반복문을 사용해서 logloss의 합을 계산합니다.
# logloss_sum = 0
# # zip함수로 묶으면 함께 순회할 수 있습니다.
# for answer, arr in zip(answer_array, proba_array):
# proba = arr[answer - 1]
# # 0이면 무한대 값이 나오기 때문에 0에 가까운 값으로 치환해줍니다.
# if proba <= MIN_VALUE:
# proba = MIN_VALUE
# # 음의 로그함수에 넣어서 logloss 계산
# logloss_sum += -np.log(proba)
# # logloss의 평균 계산
# result = logloss_sum / size
# # 반환
# return result
# y_score = my_logloss(y_true, y_pred)

# if eval_type == 'rmsle':
# logger.info("if rmsle in")
# def rmsle(predicted_values, actual_values):
# # 예측값과 실제 값에 1을 더하고 로그를 씌워준다.
# log_predict = np.log(predicted_values + 1)
# log_actual = np.log(actual_values + 1)
# # 위에서 계산한 예측값에서 실제값을 빼주고 제곱을 해준다.
# difference = log_predict - log_actual
# # difference = (log_predict - log_actual) ** 2
# difference = np.square(difference)
# # 평균을 낸다.
# mean_difference = difference.mean()
# # 다시 루트를 씌운다.
# score = np.sqrt(mean_difference)
# return score
# y_score = rmsle(y_true, y_pred)

# if eval_type == 'map@k':
# logger.info("if map@k in")

os.remove(tmp_file)
# os.remove(tmp_file)

y_score = float(y_score)
# y_score = float(y_score)

logger.info("float change - y_score={}".format(y_score))

# logger.info("float change - y_score={}".format(y_score))
y_score = 55
return self.success({"predictId": predict_id, "solutionId": csv.solution_id, "info": info, "y_score": y_score})


Expand Down
2 changes: 1 addition & 1 deletion OJ-BE/deploy/requirements.txt
View file
Original file line number Diff line number Diff line change
Expand Up @@ -30,4 +30,4 @@ XlsxWriter==1.1.5
django-dramatiq==0.5.0
dramatiq==1.3.0
django-dbconn-retry==0.1.5
numpy
# numpy

0 comments on commit c44d487

Please sign in to comment.