• The Korean Journal of Applied Statistics
• /
• v.30 no.5
• /
• pp.681-690
• /
• 2017

Various imbalanced binary classification problems exist such as fraud detection in banking operations, detecting spam mail and predicting defective products. Several sampling methods such as over sampling, under sampling, SMOTE have been developed to overcome the poor prediction performance of binary classifiers when the proportion of one group is dominant. In order to overcome this problem, several sampling methods such as over-sampling, under-sampling, SMOTE have been developed. In this study, we investigate prediction performance of logistic regression, Lasso, random forest, boosting and support vector machine in combination with the sampling methods for binary imbalanced data. Four real data sets are analyzed to see if there is a substantial improvement in prediction performance. We also emphasize some precautions when the sampling methods are implemented.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2023.05a
• /
• pp.358-358
• /
• 2023

본 연구에서는 그동안 수자원분야 강우유출 해석분야에 활용되었던 대표적인 머신러닝&딥러닝(ML&DL) 모델을 활용하여 모델의 하이퍼파라미터 튜닝뿐만 아니라 모델의 특성을 고려한 기상 및 수문데이터의 조합과 전처리(lag-time, 이동평균 등)를 통하여 데이터 특성과 ML&DL모델의 조합시나리오에 따른 일 유입량 예측성능을 비교 검토하는 연구를 수행하였다. 이를 위해 소양강댐 유역을 대상으로 1974년에서 2021년까지 축적된 기상 및 수문데이터를 활용하여 1) 강우, 2) 유입량, 3) 기상자료를 주요 영향변수(독립변수)로 고려하고, 이에 a) 지체시간(lag-time), b) 이동평균, c) 유입량의 성분분리조건을 적용하여 총 36가지 시나리오 조합을 ML&DL의 입력자료로 활용하였다. ML&DL 모델은 1) Linear Regression(LR), 2) Lasso, 3) Ridge, 4) SVR(Support Vector Regression), 5) Random Forest(RF), 6) LGBM(Light Gradient Boosting Model), 7) XGBoost의 7가지 ML방법과 8) LSTM(Long Short-Term Memory models), 9) TCN(Temporal Convolutional Network), 10) LSTM-TCN의 3가지 DL 방법, 총 10가지 ML&DL모델을 비교 검토하여 일유입량 예측을 위한 가장 적합한 데이터 조합 특성과 ML&DL모델을 성능평가와 함께 제시하였다. 학습된 모형의 유입량 예측 결과를 비교·분석한 결과, 소양강댐 유역에서는 딥러닝 중에서는 TCN모형이 가장 우수한 성능을 보였고(TCN>TCN-LSTM>LSTM), 트리기반 머신러닝중에서는 Random Forest와 LGBM이 우수한 성능을 보였으며(RF, LGBM>XGB), SVR도 LGBM수준의 우수한 성능을 나타내었다. LR, Lasso, Ridge 세가지 Regression모형은 상대적으로 낮은 성능을 보였다. 또한 소양강댐 댐유입량 예측에 대하여 강우, 유입량, 기상계열을 36가지로 조합한 결과, 입력자료에 lag-time이 적용된 강우계열의 조합 분석에서 세가지 Regression모델을 제외한 모든 모형에서 NSE(Nash-Sutcliffe Efficiency) 0.8이상(최대 0.867)의 성능을 보였으며, lag-time이 적용된 강우와 유입량계열을 조합했을 경우 NSE 0.85이상(최대 0.901)의 더 우수한 성능을 보였다.

• The Korean Journal of Applied Statistics
• /
• v.31 no.3
• /
• pp.367-382
• /
• 2018

Prediction models for a corporate bond rating in existing studies have been developed using various models such as linear regression, ordered logit, and random forest. Financial characteristics help build prediction models that are expected to be contained in the assigning model of the bond rating agencies. However, the ranges of bond ratings in existing studies vary from 5 to 20 and the prediction models were developed with samples in which the target companies and the observation periods are different. Thus, a simple comparison of the prediction accuracies in each study cannot determine the best prediction model. In order to conduct a fair comparison, this study has collected corporate bond ratings and financial characteristics from 2013 to 2017 and applied prediction models to them. In addition, we applied the elastic-net penalty for the linear regression, the ordered logit, and the ordered probit. Our comparison shows that data-driven variable selection using the elastic-net improves prediction accuracy in each corresponding model, and that the random forest is the most appropriate model in terms of prediction accuracy, which obtains 69.6% accuracy of the exact rating prediction on average from the 5-fold cross validation.

• Journal of Cadastre & Land InformatiX
• /
• v.50 no.1
• /
• pp.181-200
• /
• 2020

This study used the OLS model to estimate the determinants affecting the tenure of a home and then compared the predictive power of each model with SVM, Decision Tree, Random Forest, Gradient Boosting, XGBooest and LightGBM. There is a difference from the preceding study in that the Stacking model, one of the ensemble models, can be used as a base model to establish a more predictable model to identify the volume of housing transactions in the housing market. OLS analysis showed that sales profits, housing prices, the number of household members, and the type of residential housing (detached housing, apartments) affected the period of housing ownership, and compared the predictability of the machine learning model with RMSE, the results showed that the machine learning model had higher predictability. Afterwards, the predictive power was compared by applying each machine learning after rebuilding the data with the influencing variables, and the analysis showed the best predictive power of Random Forest. In addition, the most predictable Random Forest, Decision Tree, Gradient Boosting, and XGBooost models were applied as individual models, and the Stacking model was constructed using Linear, Ridge, and Lasso models as meta models. As a result of the analysis, the RMSE value in the Ridge model was the lowest at 0.5181, thus building the highest predictive model.

• The Korean Journal of Applied Statistics
• /
• v.30 no.1
• /
• pp.25-39
• /
• 2017

Over 5 million people participate in cycle racing betting and its revenue is more than 2 trillion won. This study predicts the ranking of cycle racing using various statistical analyses and identifies important variables which have influence on ranking. We propose competitive ranking prediction models using various classification and regression methods. Our model can predict rankings with low misclassification rates most of the time. We found that the ranking increases as the grade of a racer decreases and as overall scores increase. Inversely, we can observe that the ranking decreases when the grade of a racer increases, race number four is given, and the ranking of the last race of a racer decreases. We also found that prediction accuracy can be improved when we use centered data per race instead of raw data. However, the real profit from the future data was not high when we applied our prediction model because our model can predict only low-return events well.

• The Korean Journal of Applied Statistics
• /
• v.34 no.6
• /
• pp.863-874
• /
• 2021

In this paper, we propose a post-selection inference procedure for the fused lasso signal approximator (FLSA). The FLSA finds underlying sparse piecewise constant mean structure by applying total variation (TV) semi-norm as a penalty term. However, it is widely known that this convex relaxation can cause asymptotic inconsistency in change points detection. As a result, there can remain false change points even though we try to find the best subset of change points via a tuning procedure. To remove these false change points, we propose a post-selection inference for the FLSA. The proposed procedure applies a permutation test based on CUSUM statistic. Our post-selection inference procedure is an extension of the permutation test of Antoch and Hušková (2001) which deals with single change point problems, to multiple change points detection problems in combination with the FLSA. Numerical study results show that the proposed procedure is better than naïve z-tests and tests based on the limiting distribution of CUSUM statistics.

• The Korean Journal of Applied Statistics
• /
• v.32 no.3
• /
• pp.391-404
• /
• 2019

Customers' fixed characteristics have often been used to predict customer behavior. It has recently become possible to track customer web logs as customer activities move from offline to online. It has become possible to collect large amounts of web log data; however, the researchers only focused on organizing the log data or describing the technical characteristics. In this study, we predict the decision-making time until each customer makes the first reservation, using Airbnb customer data provided by the Kaggle website. This data set includes basic customer information such as gender, age, and web logs. We use various methodologies to find the optimal model and compare prediction errors for cases with web log data and without it. We consider six models such as Lasso, SVM, Random Forest, and XGBoost to explore the effectiveness of the web log data. As a result, we choose Random Forest as our optimal model with a misclassification rate of about 20%. In addition, we confirm that using web log data in our study doubles the prediction accuracy in predicting customer behavior compared to not using it.

• Journal of Sleep Medicine
• /
• v.15 no.2
• /
• pp.48-54
• /
• 2018

Objectives: The aim of this study was to develop a predicting model for the optimal continuous positive airway pressure (CPAP) for obstructive sleep apnea (OSA) patient with obesity by using a machine learning. Methods: We retrospectively investigated the medical records of 162 OSA patients who had obesity [body mass index (BMI) ≥ 25] and undertaken successful CPAP titration study. We divided the data to a training set (90%) and a test set (10%), randomly. We made a random forest model and a least absolute shrinkage and selection operator (lasso) regression model to predict the optimal pressure by using the training set, and then applied our models and previous reported equations to the test set. To compare the fitness of each models, we used a correlation coefficient (CC) and a mean absolute error (MAE). Results: The random forest model showed the best performance {CC 0.78 [95% confidence interval (CI) 0.43-0.93], MAE 1.20}. The lasso regression model also showed the improved result [CC 0.78 (95% CI 0.42-0.93), MAE 1.26] compared to the Hoffstein equation [CC 0.68 (95% CI 0.23-0.89), MAE 1.34] and the Choi's equation [CC 0.72 (95% CI 0.30-0.90), MAE 1.40]. Conclusions: Our random forest model and lasso model ($26.213+0.084{\times}BMI+0.004{\times}$apnea-hypopnea index+$0.004{\times}oxygen$ desaturation index-$0.215{\times}mean$ oxygen saturation) showed the improved performance compared to the previous reported equations. The further study for other subgroup or phenotype of OSA is required.

• Journal of the Korean Data and Information Science Society
• /
• v.24 no.1
• /
• pp.125-133
• /
• 2013

In this paper, penalized binary logistic regression models are employed as statistical models for determining the discharge of 668 patients with a chief complaint of dyspnea based on 11 blood tests results. Specifically, the ridge model based on $L^2$ penalty and the Lasso model based on $L^1$ penalty are considered in this paper. In the comparison of prediction accuracy, our models are compared with the logistic regression models with all 11 explanatory variables and the selected variables by variable selection method. The results show that the prediction accuracy of the ridge logistic regression model is the best among 4 models based on 10-fold cross-validation.

• The Korean Journal of Applied Statistics
• /
• v.26 no.3
• /
• pp.441-452
• /
• 2013

This study analyzed the relationship between key film and a box office record success factors based on movies released in the first quarter of 2013 in Korea. An over-fitting problem can happen if there are too many explanatory variables inserted to regression model; in addition, there is a risk that the estimator is instable when there is multi-collinearity among the explanatory variables. For this reason, optimal variable selection based on high explanatory variables in box-office performance is of importance. Among the numerous ways to select variables, LASSO estimation applied by a generalized linear model has the smallest prediction error that can efficiently and quickly find variables with the highest explanatory power to box-office performance in order.