Earticle

4차산업혁명의 핵심인 인공지능 기술은 인간의 능력을 뛰어넘어 주식예측에도 적용하고 있으면 예측이 불가능 한 것을 딥러닝 기법과 머신러닝을 통하여 지능화된 판단을 내리고 있는 실정이다. 미국의 펀드매니지먼트 회사에서는 증시 에널리스트의 역할을 인공지능이 대신하고 있으며, 이 분야의 연구가 활발히 진행 중에 있다. 본 연구에서는 BLSTM을 이용하여 기존의 LSTM방식의 단방향 예측에서 발생하는 오류를 줄이고, 양방향으로 예측하여 예측에 대한 오류를 줄이고, 주식 가격에 영향을 미치는 거시 지표, 즉 경제성장률, 경제지표, 이자율, 무역수지, 환율, 통화량을 분 석한다. 거시 지표 분석 후에 개별 주식에 대한 PBR, BPS, ROE 예측과 가장 주식 가격에 영향을 미치는 외국인, 기관, 연기금 등 매수와 매도 물량을 분석하여 주식의 목표주가를 정확히 예측하여 주식 투자에 도움을 주기 위해 본 연구를 수행했다.

Artificial intelligence technology, which is the core of the 4th industrial revolution, is making intelligent judgments through deep learning techniques and machine learning that it is impossible to predict if it is applied to stock prediction beyond human capabilities. In US fund management companies, artificial intelligence is replacing the role of stock market analyst, and research in this field is actively underway. In this study, we use BLSTM to reduce errors that occur in unidirectional prediction of the existing LSTM method, reduce errors in predictions by predicting in both directions, and macroscopic indicators that affect stock prices, namely, economic growth rate, economic indicators, interest rate, analyze the trade balance, exchange rate, and volume of currency. To help stock investment by accurately predicting the target price of stocks by analyzing the PBR, BPS, and ROE of individual stocks after analyzing macro-indicators, and by analyzing the purchase and sale quantities of foreigners, institutions, pension funds, etc., which have the most influence on stock prices.

본 연구는 2014 시즌의 한국여자프로골프선수의 상금 순위와 관련된 경기력을 분석하고, 분석된 결과를 사용하여특정한 경기력 요인의 변화에 대한 상금 순위를 예측하는 시스템을 설계하고자 한다. 경기력 분석을 위한 골프경기력 요인으로 평균타수, 평균퍼팅, 평균버디율, 그린적중률, 파세이브율, 파브레이크율, 리커버리율을 사용하였다. 각각의 경기력 요인은 곡선맞춤함수를 이용한 비선형회귀분석을 하였으며, 이는 각각의 경기력 요인에 대해예상순위를 알 수 있다. 따라서 특정한 선수에 있어서 각 경기력 요인에 대한 현재순위와 예상순위를 비교하여강점과 약점의 분석이 가능하다. 특히, 취약한 경기력 요인이 보완될 경우, 다른 경기력 요인의 예측과 분석이 가능하도록 시스템을 설계하였다. 본 연구에서는 특정한 선수의 취약한 경기력 요인이 개선될 경우, 설계된 시스템의 예측 결과와 2015 시즌의 실제 경기력과 비교하여 분석하였다. 분석된 결과 예측된 시스템의 결과와 2015 시즌의 경기력과 유사함을 알 수 있었으며, 상금 순위를 예측하는 시스템의 설계가 타당함을 알 수 있었다.

This study is to analyze KLPGA player's performance of the 2014 season related to the prize money and design a prediction system of it according to the alternation of the performance factors based on the formerly analyzed facts. For the performance factors of the performance analysis, this study used scoring average, putting average, average birdies, green in regulation, par saves, par breakers, and recovery rate. Each of these have been completed with nonlinear regression analysis using curve-fitting function, which enabled this study to figure out each of the prediction ranking for the performance factor. Therefore, it was possible to examine the strengths and weaknesses of a particular player by comparing present ranking and expectation ranking on each of the performance factor. Particularly, when vulnerable performance factor is complemented, the system is designed to predict and analyze other performance factors. In this study, when vulnerable performance factor of a particular player was improved, the analysis between the designed system's prediction outcomes and 2015 season's real performances was done. To conclude, the system's prediction outcomes and 2015 season's performances were similar. Because of that it would be possible to say that the design of the prize money prediction system is valid.

Crowdsensing technologies can improve the efficiency of smart parking system in comparison with present sensor based smart parking system because of low install price and no restriction caused by sensor installation. A lot of sensing data is necessary to predict parking lot saturation in real-time. However in real world, it is hard to reach the required number of sensing data. In this paper, we model a saturation predication combining a time-based prediction model and a sensing data-based prediction model. The time-based model predicts saturation in aspects of parking lot location and time. The sensing data-based model predicts the degree of saturation of the parking lot with high accuracy based on the degree of saturation predicted from the first model, the saturation information in the sensing data, and the number of parking spaces in the sensing data. We perform prediction model learning with real sensing data gathered from a specific parking lot. We also evaluate the performance of the predictive model and show its efficiency and feasibility.

A wastewater treatment plant is an intricate system with a wealth of information, where the aeration system of the active sludge process is designed to provide oxygen to microorganisms. Owing to the time delay in biochemical reactions, adjustments made by operational staff to the airflow often lead to delayed changes in dissolved oxygen concentration, frequently causing overaeration. The paper introduces a machine learning model that utilizes water quality indicators and air blower indicators to predict current airflow. By leveraging the airflow predicted by this model, the dissolved oxygen concentration for the next hour is successfully maintained within the optimal range of 2 mg/L to 4 mg/L. In the case of airflow prediction, the Transformer model proved more effective than the random forest and long short-term memory models, owing to its self-attention model architecture. In conclusion, the study demonstrates the successful applicability of machine learning models to predict airflow on the promise of maintaining dissolved oxygen stability. These findings present a data-driven approach to guarantee the steady operation of wastewater treatment plants.

With the rapid growth of the Internet, the Internet-based robot has been realized by connecting off-line robot to the Internet. However, because the Internet is often irregular and unreliable, the varying time delay in data transmission is a significant problem for the construction of the Internet-based robot system. Thus, this paper is concerned with the development of an Internet-based robot system, which is insensitive to the Internet time delay. For this purpose, the PPS (Position Prediction Simulator) is suggested and implemented on the system. The PPS consists of two parts : the robot position prediction part and the projective virtual scene part. In the robot position prediction part, the robot position is predicted for more accurate operation of the mobile robot, based on the time at which the user's command reaches the robot system. The projective virtual scene part shows the 3D visual information of a remote site, which is obtained through image processing and position prediction. For the verification of this proposed PPS, the robot was moved to follow the planned path under the various network traffic conditions. The simulation and experimental results showed that the path error of the robot motion could be reduced using the developed PPS.

An early warning system was developed to help identify stock status as early as possible. For performance to improve, there needs to be a feature to predict the amount of stock that must be provided and a feature to estimate when to buy goods. This research was conducted to improve the inventory early warning system and optimize the Reminder Block's performance in minimum stock settings. The models used in this study are the single exponential smoothing (SES) method for prediction and the economic order quantity (EOQ) model for determining the quantity. The research was conducted by analyzing the Reminder Block in the early warning system, identifying data needs, and implementing the SES and EOQ mathematical models into the Reminder Block. This research proposes a new Reminder Block that has been added to the SES and EOQ models. It is hoped that this study will help in obtaining accurate information about the time and quantity of repurchases for efficient inventory management.

The gas metal arc (GMA) welding is one of the most widely-used processes in metal joining process that involves the melting and solidification of the joined materials. To solve this problem, we have carried out the sequential experiment based on a Taguchi method and identified the various problems that result from the robotic GMA welding process to characterize the GMA welding process and establish guidelines for the most effective joint design. Also using multiple regression analysis with the help of a standard statistical package program, SPSS, on an IBM-compatible PC, three empirical models (linear, interaction, quadratic model) have been developed for off-line control which studies the influence of welding parameters on bead width and compares their influences on the bead width to check which process parameter is most affecting. These models developed have been employed for the prediction of optimal welding parameters and assisted in the generation of process control algorithms.

Purpose: The purpose of this study was to evaluate the relation between PBS scores and GMFCS levels and to examine whether pediatric balance scale (PBS) scores were useful for predicting gross motor functional classification system (GMFCS) levels in children with cerebral palsy. Methods: This cross-sectional study was performed conducted for to evaluatione of PBS and GMFCS using in 26 children with cerebral palsy (16 males and 10 females with GMFCS level I to III). PBS total and item scores at different levels of GMFCS were measured. Results: The hHigh PBS item average scores obtained from standing and postural change dimensions except sitting dimension were observed at the low levels of GMFCS and these results were statistically significant (p<0.05). The relation between PBS (standing and postural change dimensions) and GMFCS levels were was significantly different, except the relation between PBS sitting dimension and GMFCS levels showing a ceiling effect. Conclusion: GMFCS is designed to for classificationy of gross motor functions emphasizing on walking movement and PBS is was developed to for evaluatione of functional balance. Based on the results of this study showing high relation between GMFCS levels and PBS scores, PBS scores can be used for predicting GMFCS levels.

In this paper, a prediction system is proposed to control the brightness of smart street lamps by predicting the moving path through the reduction of consumption power and information of pedestrian’s past moving direction while meeting the function of existing smart street lamps. The brightness of smart street lamps is adjusted by utilizing the walk tracking vector and soft hand-off characteristics obtained through the motion sensing sensor of smart street lamps. In addition, the motion vector is used to analyze and predict the pedestrian path, and the GPU is used for high-speed computation. Pedestrians were detected using adaptive Gaussian mixing, weighted difference imaging, and motion vectors, and motions of pedestrians were analyzed using the extracted motion vectors. The preprocessing process using linear interpolation is performed to improve the performance of the proposed prediction system. Fuzzy prediction system and neural network prediction system are designed in parallel to improve efficiency and rough set is used for error correction.

중국 우한발 코로나 19 바이러스로 인하여 세계 경제가 침체하여, 미국연방준비제도를 비롯한 대부분 국가에서 는 통화량을 늘려 경기를 부양하는 정책을 내놓았다. 주식 투자자들 대부분은 기업에 대한 재무제표 분석이 없이 유명 유튜버의 추천종목이나 지인의 말만 듣고 투자하는 경향이 있어서 주식투자의 손실 가능성이 크다. 따라서, 본 연구에 서는 기존 자동매매 조건에서 발전된 인공지능 딥러닝 기법을 이용하여 주가에 영향을 미치는 거시지표를 분석하고 예측하여 주가에 미치는 상관관계를 통한 개별주가예측에 가중치를 부여하고 주가를 예측한다. 또한, 주가는 실시간 증시뉴스에 민감하게 반응하기 때문에 증시뉴스 텍스트 마이닝을 통하여 인공지능으로 예측된 주가에 가중치를 반영하 여 더 정확한 주가 예측을 하여 주식 투자자에게 매매의 판단 근거를 제공하여 건전한 주식투자가 되도록 이바지하였다.

As the global economy stagnated due to the Corona 19 virus from Wuhan, China, most countries, including the US Federal Reserve System, introduced policies to boost the economy by increasing the amount of money. Most of the stock investors tend to invest only by listening to the recommendations of famous YouTubers or acquaintances without analyzing the financial statements of the company, so there is a high possibility of the loss of stock investments. Therefore, in this research, I have used artificial intelligence deep learning techniques developed under the existing automatic trading conditions to analyze and predict macro-indicators that affect stock prices, giving weights on individual stock price predictions through correlations that affect stock prices. In addition, since stock prices react sensitively to real-time stock market news, a more accurate stock price prediction is made by reflecting the weight to the stock price predicted by artificial intelligence through stock market news text mining, providing stock investors with the basis for deciding to make a proper stock investment.

주가는 사람들의 심리를 반영하고 있으며, 주식시장 전체에 영향을 미치는 요인으로는 경제성장률, 경제지료, 이자율, 무역수지, 환율, 통화량 등이 있다. 국내 주식시장은 전날 미국 및 주변 국가들의 주가지수에 영향을 많이 받고 있으며 대표적인 주가지수가 다우지수, 나스닥, S&P500이다. 최근 주가뉴스를 이용한 주가분석 연구가 활발히 진행되 고 있으며, 인공지능 기반한 분석을 통하여 과거 시계열 데이터를 기반으로 미래를 예측하는 연구가 진행 중에 있다. 하지만, 주식시장은 예측시스템에 의해서 단기간 적중이 되더라도, 시장은 더 이상의 단기 전략대로 움직여지지 않고, 새롭게 변할 수밖에 없다. 따라서, 본 모델을 삼성전자 주식데이터와 뉴스 정보를 텍스트 마이닝으로 모니터링하여 분 석한 결과를 나타내어 예측이 가능한 모델을 제시하였으며, 향후 종목별 예측을 통하여 실제 예측이 정확한지 확인하여 발전시켜 나갈 예정임

The stock price reflects people's psychology, and factors affecting the entire stock market include economic growth rate, economic rate, interest rate, trade balance, exchange rate, and currency. The domestic stock market is heavily influenced by the stock index of the United States and neighboring countries on the previous day, and the representative stock indexes are the Dow index, NASDAQ, and S & P500. Recently, research on stock price analysis using stock news has been actively conducted, and research is underway to predict the future based on past time series data through artificial intelligence-based analysis. However, even if the stock market is hit for a short period of time by the forecasting system, the market will no longer move according to the short-term strategy, and it will have to change anew. Therefore, this model monitored Samsung Electronics' stock data and news information through text mining, and presented a predictable model by showing the analyzed results.

The purpose of this study is to support the decision of semiconductor companies by providing an objective chip price prediction model. Existing statistical or econometric models have shown limits analyzing nonlinear time-series data such as share prices and exchange rates. The back-propagation algorithm, which is the most common method, was used as the learning algorithm. Predicting was attempted by using two supply factor variables and four demand factor variables. The data used in the analysis was collected from January 3, 2003 to December 28, 2005. The data has been divided into two parts for learning and verification. As a result of inputting the verification data into the trained neural network, the actual values show some differences. However, we were able to see that the flow of the semiconductor market and short-term forecasting was possible providing very little error between the predicted value and the actual value.

This study intends to link agricultural machine history data with related organizations or collect them through IoT sensors, receive input from agricultural machine users and managers, and analyze them through AI algorithms. Through this, the goal is to track and manage the history data throughout all stages of production, purchase, operation, and disposal of agricultural machinery. First, LSTM (Long Short-Term Memory) is used to estimate oil consumption and recommend maintenance from historical data of agricultural machines such as tractors and combines, and C-LSTM (Convolution Long Short-Term Memory) is used to diagnose and determine failures. Memory) to build a deep learning algorithm. Second, in order to collect historical data of agricultural machinery, IoT sensors including GPS module, gyro sensor, acceleration sensor, and temperature and humidity sensor are attached to agricultural machinery to automatically collect data. Third, event-type data such as agricultural machine production, purchase, and disposal are automatically collected from related organizations to design an interface that can integrate the entire life cycle history data and collect data through this.

In this paper, we propose a system for controlling the brightness of street lights by predicting pedestrian paths, identifying the position of pedestrians with motion sensing sensors and obtaining motion vectors based on past walking directions, then predicting pedestrian paths through the route prediction smart street lighting system. In addition, by using motion vector data, the pre-treatment process using linear interpolation method and the fuzzy system and neural network system were designed in parallel structure to increase efficiency and the rough set was used to correct errors. It is expected that the system proposed in this paper will be effective in securing the safety of pedestrians and reducing light pollution and energy by predicting the path of pedestrians in the detection of movement of pedestrians and in conjunction with smart street lightings.

This study proposes an AI system that predicts customer churn based on telecommunications customer data and automatically generates personalized retention strategies for each customer. Recently, due to market saturation and intensified competition in the telecommunications industry, retaining existing customers has become more important than acquiring new ones. As a result, predicting customer churn in advance and responding effectively has become a key factor in maintaining a company’s competitiveness. However, the causes of customer churn are diverse, including dissatisfaction with pricing, service quality issues, lack of benefits, and customer service experiences. In particular, much of this information exists in the form of unstructured text data such as consultation records and VOC (Voice of Customer), making systematic analysis difficult. To address this issue, this study utilizes natural language processing (NLP) techniques to analyze the reasons for customer churn and convert them into quantitative measures. Specifically, deep learning-based language models such as DistilBERT and RoBERTa were used to process customer text data and perform sentiment analysis. Based on the results, a churn_score representing the customer’s churn risk was calculated. This approach enables the transformation of previously qualitative customer dissatisfaction factors into quantitative indicators for analysis. In addition, this study employs the LightGBM algorithm for structured data-based churn prediction. This model considers various factors such as customer pricing plans, usage patterns, and contract duration to predict churn probability, offering both high predictive performance and efficiency. Furthermore, beyond simply predicting churn, this study designs a system that automatically generates customized retention strategies tailored to each customer based on the analysis results. This allows companies to establish more timely and appropriate response strategies for individual customers. In conclusion, this study proposes an integrated approach that combines unstructured text data and structured data to analyze and predict customer churn, while also providing actionable retention strategies. This system can significantly enhance data-driven customer retention strategies in the telecommunications industry and is expected to be applicable to various other industries in the future.

To better predict and classify failures of information system development projects (ISDPs), this study proposes ISDPs failure prediction models using the traditional statistical methods and artificial intelligence methods, namely multiple discriminant analysis (MDA), logistic regression (LR), multi-layer perceptron (MLP), classification and regression tree (CART), commercial version 5.0 (C5.0), and support vector machine (SVM). We performed the analysis on the audit report data of 446 projects that were conducted by a global information technology (IT) company, to build the IT service systems and relevant service infrastructures needed for a project with South Korea’s mobile telecommunication companies. The research variables, which were confirmed by project performance management system (PPMS) of the company, are composed of thirteen variables. Empirical results indicated that SVM outperforms other models such as MDA, LR, MLP, CART, and C5.0.