• The Korean Journal of Applied Statistics
• /
• v.34 no.5
• /
• pp.773-794
• /
• 2021

Prediction of vehicle traffic volume is very important in planning municipal administration. It may help promote social and economic interests and also prevent traffic congestion costs. Traffic volume as a time-varying trajectory is considered as functional data. In this paper we study three functional regression models that can be used to predict an unseen trajectory of traffic volume based on already observed trajectories. We apply the methods to highway tollgate traffic volume data collected at some tollgates in Seoul, Chuncheon and Gangneung. We compare the prediction errors of the three models to find the best one for each of the three tollgate traffic volumes.

• The Korean Journal of Applied Statistics
• /
• v.34 no.3
• /
• pp.329-345
• /
• 2021

Recommender systems use data from customers to suggest personalized products. The recommender systems can be categorized into three cases; collaborative filtering, contents-based filtering, and hybrid recommender system that combines the first two filtering methods. In this work, we introduce and compare deep learning-based recommender system using autoencoder. Autoencoder is an unsupervised deep learning that can effective solve the problem of sparsity in the data matrix. Five versions of autoencoder-based deep learning models are compared via three real data sets. The first three methods are collaborative filtering and the others are hybrid methods. The data sets are composed of customers' ratings having integer values from one to five. The three data sets are sparse data matrix with many zeroes due to non-responses.

• The Korean Journal of Applied Statistics
• /
• v.34 no.2
• /
• pp.141-152
• /
• 2021

A new statistical model needs additional variables in order to re-evaluate the undecided inference. Then the MNAR assumption is required, since the probabilities for the positivity of the indeterminant and the determinant is calculated differently. In this study, since two statistical models have a hierarchical relationship, we determine the undecided inference under the MNAR assumption using the confidence interval of the difference between two AUCs. Among many methods of estimating the confidence interval of the AUC difference, it is found that four kinds of methods show excellent performance through simulations. And based on these methods, we propose a variable selection method that are useful for the undecided inference using logistic regression models.

• The Korean Journal of Applied Statistics
• /
• v.34 no.2
• /
• pp.177-190
• /
• 2021

Random forests is a popular method that improves the instability and accuracy of decision trees by ensembles. In contrast to increasing the accuracy, the ease of interpretation is sacrificed; hence, to compensate for this, variable importance is provided. The variable importance indicates which variable plays a role more importantly in constructing the random forests. However, when a predictor is correlated with other predictors, the variable importance of the existing importance algorithm may be distorted. The downward bias of correlated predictors may reduce the importance of truly important predictors. We propose a new algorithm remedying the downward bias of correlated predictors. The performance of the proposed algorithm is demonstrated by the simulated data and illustrated by the real data.

• The Korean Journal of Applied Statistics
• /
• v.34 no.1
• /
• pp.9-23
• /
• 2021

In the following paper we introduce a variational Bayes method that approximates posterior distributions with mean-field method. In particular, we introduce automatic differentiation variation inference (ADVI), which approximates joint posterior distributions using the product of Gaussian distributions after transforming parameters into real coordinate space, and then apply it to pharmacokinetic models that are models for the study of the time course of drug absorption, distribution, metabolism and excretion. We analyze real data sets using ADVI and compare the results with those based on Markov chain Monte Carlo. We implement the algorithms using Stan.

• The Korean Journal of Applied Statistics
• /
• v.33 no.6
• /
• pp.665-682
• /
• 2020

This paper introduces numba library in Python, which improves computational efficiency of the provided implemented code written by naive Python language by applying just-in-time (JIT) compilation. To apply just-in-time compilation, the numba only needs to use a decorator on a target Python function. We provide implementation examples with numba for the permutation test and the parameter estimation for Gaussian mixture distribution. We also numerically show the efficiency of numba by comparing the total computation times of the implementation using naive python and the implementation using numba for each application.

• The Journal of the Convergence on Culture Technology
• /
• v.8 no.1
• /
• pp.531-536
• /
• 2022

This paper compared the performance of the parametric method and the nonparametric method when estimating the distribution for the tail of the distribution with heavy tails. For the parametric method, the generalized extreme value distribution and the generalized Pareto distribution were used, and for the nonparametric method, the kernel density estimation method was applied. For comparison of the two approaches, the results of function estimation by applying the block maximum value model and the threshold excess model using daily fine dust public data for each observatory in Seoul from 2014 to 2018 are shown together. In addition, the area where high concentrations of fine dust will occur was predicted through the return level.

• The Korean Journal of Applied Statistics
• /
• v.35 no.1
• /
• pp.147-159
• /
• 2022

Nowadays, various social media platforms are widely spread and people closely use such platforms in daily life. By doing so, social influencers with a large number of subscribers, views, and comments have huge impact in our society. Following this trend, many companies are actively using influencers for marketing purpose to promote their products and services. In this study, we extract the speeches of influencers from videos for game marketing and analyze them using various text mining techniques. In the analysis, we distinguish game videos leading to successful marketing and failed marketing, and we explore and compare the linguistic features of the influencers for successful and failed marketings.

• The Korean Journal of Applied Statistics
• /
• v.35 no.3
• /
• pp.445-455
• /
• 2022

In this paper, we introduce existing Bayesian methods for high-dimensional sparse regression models and compare their performance in various simulation scenarios. Especially, we focus on the variational Bayes approach proposed by Ray and Szabó (2021), which enables scalable and accurate Bayesian inference. Based on simulated data sets from sparse high-dimensional linear regression models, we compare the variational Bayes approach with other Bayesian and frequentist methods. To check the practical performance of the variational Bayes in logistic regression models, a real data analysis is conducted using leukemia data set.

• Journal of the Korean Society of Industry Convergence
• /
• v.24 no.3
• /
• pp.323-331
• /
• 2021

With global warming and pollution problems, accurate forecasting of the harmful gases would be an essential alarm in our life. In this paper, we forecast the emission of the five gases(SOx, NO₂, NH₃, H₂S, CH₄) using the time series model of ARIMA, the learning algorithms of Random forest, and LSTM. We find that the gas emission data depends on the short-term memory and behaves like a random walk. As a result, we compare the RMSE, MAE, and MAPE as the measure of the prediction performance under the same conditions given to three models. We find that ARIMA forecasts the gas emissions more precisely than the other two learning-based methods. Besides, the ARIMA model is more suitable for the real-time forecasts of gas emissions because it is faster for modeling than the two learning algorithms.