수요 예측 평가를 위한 가중절대누적오차지표의 개발

최대일,옥창수 한국산업경영시스템학회 2015 한국산업경영시스템학회지 Vol.38 No.3

Aggregate Production Planning determines levels of production, human resources, inventory to maximize company’s profits and fulfill customer's demands based on demand forecasts. Since performance of aggregate production planning heavily depends on accuracy of given forecasting demands, choosing an accurate forecasting method should be antecedent for achieving a good aggregate production planning. Generally, typical forecasting error metrics such as MSE (Mean Squared Error), MAD (Mean Absolute Deviation), MAPE (Mean Absolute Percentage Error), and CFE (Cumulated Forecast Error) are utilized to choose a proper forecasting method for an aggregate production planning. However, these metrics are designed only to measure a difference between real and forecast demands and they are not able to consider any results such as increasing cost or decreasing profit caused by forecasting error. Consequently, the traditional metrics fail to give enough explanation to select a good forecasting method in aggregate production planning. To overcome this limitation of typical metrics for forecasting method this study suggests a new metric, WACFE (Weighted Absolute and Cumulative Forecast Error), to evaluate forecasting methods. Basically, the WACFE is designed to consider not only forecasting errors but also costs which the errors might cause in for Aggregate Production Planning. The WACFE is a product sum of cumulative forecasting error and weight factors for backorder and inventory costs. We demonstrate the effectiveness of the proposed metric by conducting intensive experiments with demand data sets from M3-competition. Finally, we showed that the WACFE provides a higher correlation with the total cost than other metrics and, consequently, is a better performance in selection of forecasting methods for aggregate production planning.

한국의 해양예측, 오늘과 내일

김영호,최병주,이준수,변도성,강기룡,김영규,조양기,Kim, Young Ho,Choi, Byoung-Ju,Lee, Jun-Soo,Byun, Do-Seong,Kang, Kiryong,Kim, Young-Gyu,Cho, Yang-Ki 한국해양학회 2013 바다 Vol.18 No.2

경제 발전에 따라 레저, 해운, 수산, 국방, 해난사고 등 해양을 이용하는 활동이 증가하면서 해양예보에 대한 수요가 크게 증가하고 있다. 기상에서 해양의 역할이 새롭게 인식되면서 정확한 기상 및 기후변화를 예측하기 위한 해양 예측의 필요성도 증가하고 있다. 사회적인 요구와 관련 기술의 발전에 힘입어 선진국을 중심으로 해양예측시스템이 수립되어 왔다. 이 연구에서는 세계적으로 해양예측시스템을 발전시키고 확산시킨 국제협력프로그램 GODAE(Global Ocean Data Assimilation Experiment)의 진행과정과 기여를 정리하였다. 그리고 현재 해양예측시스템을 운용 중인 미국, 프랑스, 영국, 이탈리아, 노르웨이, 호주, 일본, 중국이 해양예측시스템을 구축하면서 세웠던 목적과 비전, 역사, 연구 동향을 조사하고 각 나라의 해양예측시스템 현황을 비교하였다. 우리보다 앞서 해양예측시스템을 구축하여 사용하고 있는 나라들이 취한 개발 전략의 특징은 다음과 같이 요약해 볼 수 있다. 첫째, 국가적인 역량을 집중하여 성공적인 현업 해양예측시스템을 구축하였다. 둘째, 국제적인 프로그램을 통해 선진 기술을 공유하고 상호 발전시켰다. 셋째, 각 기관의 역할과 고유 목적에 따라 기여분야를 나눠가졌다. 국내에서도 최근 현업 해양예측시스템에 대한 수요가 증대되고 있다. 기상청, 국립해양조사원, 국립수산과학원, 국방과학연구소의 해양예측시스템 개발에 관한 현재 상황과 향후 장기적 계획을 조사하였다. 국지 해양예측 또는 기후예측 모델을 위한 개방경계 초기장 제공이 가능한 광역의 정확도 높은 해양예측시스템을 구축하기 위해서는 국내의 유관 기관 간 협력 관계가 필수적이다. 이를 위해 관련 기관과 연구자들이 함께 참여하는 컨소시엄 형성이 바람직하다. 컨소시엄을 통해 경쟁력 높은 예측 모델과 시스템을 구축할 수 있으며, 제한된 재원을 효율적으로 활용할 수 있고, 연구 개발 인력이 전문분야에 집중할 수 있으며, 중복 투자를 막고 각 기관은 고유 업무에 역량을 집중할 수 있다. 비록 해양예보에 있어 우리나라가 현 단계로는 국제적인 수준에 뒤쳐져 있지만, 각 유관 기관들이 고유 업무를 정립하고 국가적인 역량을 집중하여 현업 해양예측시스템을 공동 개발하면 곧 추격하여 해양예보 분야를 선도할 수 있을 것이다. National demands for the ocean forecasting system have been increased to support economic activity and national safety including search and rescue, maritime defense, fisheries, port management, leisure activities and marine transportation. Further, the ocean forecasting has been regarded as one of the key components to improve the weather and climate forecasting. Due to the national demands as well as improvement of the technology, the ocean forecasting systems have been established among advanced countries since late 1990. Global Ocean Data Assimilation Experiment (GODAE) significantly contributed to the achievement and world-wide spreading of ocean forecasting systems. Four stages of GODAE were summarized. Goal, vision, development history and research on ocean forecasting system of the advanced countries such as USA, France, UK, Italy, Norway, Australia, Japan, China, who operationally use the systems, were examined and compared. Strategies of the successfully established ocean forecasting systems can be summarized as follows: First, concentration of the national ability is required to establish successful operational ocean forecasting system. Second, newly developed technologies were shared with other countries and they achieved mutual and cooperative development through the international program. Third, each participating organization has devoted to its own task according to its role. In Korean society, demands on the ocean forecasting system have been also extended. Present status on development of the ocean forecasting system and long-term plan of KMA (Korea Meteorological Administration), KHOA (Korea Hydrographic and Oceanographic Administration), NFRDI (National Fisheries Research & Development Institute), ADD (Agency for Defense Development) were surveyed. From the history of the pre-established systems in other countries, the cooperation among the relevant Korean organizations is essential to establish the accurate and successful ocean forecasting system, and they can form a consortium. Through the cooperation, we can (1) set up high-quality ocean forecasting models and systems, (2) efficiently invest and distribute financial resources without duplicate investment, (3) overcome lack of manpower for the development. At present stage, it is strongly requested to concentrate national resources on developing a large-scale operational Korea Ocean Forecasting System which can produce open boundary and initial conditions for local ocean and climate forecasting models. Once the system is established, each organization can modify the system for its own specialized purpose. In addition, we can contribute to the international ocean prediction community.

머신러닝과 시계열 기법 기반의 초단기 시간단위 수요예측방법론 개발 연구

민경창,하헌구 한국로지스틱스학회 2022 로지스틱스연구 Vol.30 No.3

Demand forecasting is an important field and it is safe to say that forecasting is a key component of economic activity. An accurate forecasting is the key to determining the competitiveness of all economic players. Forecasting an uncertain future is a difficult task and radical change in the external environment are adding to the difficulty of forecasting. Amid the increasing demand for accurate demand forecasting, the emergence of Big data, AI, ML, and DL following the development of computing power is becoming a major turning point in the demand forecasting field as well. In addition to the traditional forecasting methodologies, the use of dataming techniques is also rapidly increasing. And various efforts have been continued to improve the forecasting accuracy. In this paper, a hybrid forecasting methodology which is combined time series model and data mining technique and a multistage methodology are presented for short-term forecasting. Specifically, we developed a hybrid forecasting model that combines SARIMA(Seasonal Autoregressive Integrated Moving Average) and Random Forest, and a multistage methodology that utilizing the forecasting result of the upper-category as a variable in the forecasting process of the sub-category. In order to verify the methodologies presented in this paper, we use the rental data of ‘Seoul bike’(shared bicycle in Seoul) as verification data. As a result of the forecasting ‘Seoul bike’ demand for the next 7 days(every 3 hours) of rental point clusters, the average forecasting accuracy was 81.5%. It is high accuracy level considering that the forecasting unit was 3hours, forecasting horizon was next 56 steps, and the average accuracy by Random forest was 65%. In addition, it was confirmed that high accuracy was maintained steadily regardless of the time difference from the forecasting point unlike the characteristics of general demand forecasting, And the high accuracy level was confirmed as a forecasting model not only a 3 hours forecasting, but also daily(90.1%) and weekly(91.7%) forecasting. The research shows the forecasting methodologies of this paper is worth to use as a short-term forecasting model. And we confirmed that the methodologies are very useful to forecasting daily and weekly demand as well. It is expected that the methodologies proposed in this paper will be widely used as an accurate forecasting model in more diverse fields.

Temporal Fusion Transformer를 이용한 대형마트 판매량의 다단계 시계열 수요예측

안세희,정재윤 한국전자거래학회 2023 한국전자거래학회지 Vol.28 No.3

Demand forecast is used as basic data for business and operation planning in all industries. In this paper, the Temporal Fusion Transformer (TFT) architecture was applied to the data of the M5 Competition, a famous forecasting competition, and the accuracy of the TFT-based forecasting method was compared with that of the DRFAM method, that had won the competition. The performance was evaluated for the sales data of CA_1 store in the Walmart dataset of the M5 Competition. The TFT models were trained with two data pools at the store level and category level, respectively, and the final forecast was calculated by arithmetically averaging the prediction results of the two models. As a result, the TFT-based method obtained better forecasts than the DRFAM method, which trained six LightGBM models with direct forecasting and recursive forecasting for three levels of data pools and predicted with the arithmetic average of the six trained models. It was found that the TFT-based method had sufficiently learned the relationship between variables and sales volumes in the time-series using the self-attention structure of TFT. While the direct and recursive forecasting models of the DRFAM method require 28 repeated calls for 28 days of forecasting, the TFT-based method can obtain 28 time-series forecasts with a single model call because of its multi-output structure. The proposed TFT-based forecasting method is expected to be applicable to various fields by providing faster and more accurate time-series forecasts. 수요예측은 모든 산업에서 사업 기획 및 운영 계획의 중요한 기초 자료로 사용된다. 본 논문에서는 수요예측 경진대회인 M5 Competition 데이터를 대상으로 Temporal Fusion Transformer(TFT) 모형을 적용하였고, 이 대회에서 우승한 DRFAM 기법과 정확도를 비교하였다. M5 Competition의 Walmart 데이터셋 중 CA_1 매장의 판매량 데이터를 대상으로 성능을 평가하였으며, 매장(store) 수준과 카테고리(category) 수준의 데이터풀(data pool)로 각각 TFT 모형을 학습한 후 예측값을 산술평균하는 방식을 사용하였다. 그 결과, 세 가지 수준의 데이터풀에 대해 직접적 예측모형(direct forecasting)과 재귀적 예측모형(recursive forecasting)으로 총 6개의 LightGBM 모형을 학습하여 산술평균으로 예측하는 DRFAM 기법보다 평균적으로 개선된 예측 정확도를 달성하였다. 이를 통해 TFT 모형이 자기-어텐션 구조를 사용하여 시계열에서 변수와 판매량 간의 관계를 충분히 학습하였음을 알 수 있었다. DRFAM 기법의 직접적 예측모형과 재귀적 예측모형이 28일 간의 예측을 위하여 28회 반복호출을 해야 하지만, TFT 모형은 다중 출력 구조이기 때문에 한번 모형 호출로 28개의 시계열 예측이 가능하다. 본 논문에서 제안한 TFT 기반의 예측모형은 보다 빠르고 정확한 시계열 예측을 제공하여 다양한 분야에 확대 적용할 수 있을 것으로 기대한다.

재무분석가의 현금흐름예측 활동이 정보비대칭에 미치는 영향

인창열,김태희,이명곤 한국공인회계사회 2017 회계·세무와 감사 연구 Vol.59 No.2

The purpose of this study is to examine whether information asymmetry decreases as financial analysts’ cash flow forecasting activities increase. Concretely, this study analyses whether information asymmetry is relieved according to financial analysts’ provision of cash flow forecasts, cash flow forecasting frequency, and the ratio of earnings forecasting frequency to cash flow forecasting frequency(hereinafter, ‘cash flow forecasting ratio’). In addition, this study examines whether differences in the relationship between cash flow forecasting activities and information asymmetry exist according to the quality of earnings. Financial analysts play the role of mitigating information asymmetry in the capital market by providing useful forecast information to capital market participants based on superior abilities and expert knowledge. When financial analysts provide cash flow forecasts, various information such as cash flow forecasts are provided to capital market participants. In other words, analysts’ cash flow forecasts are expected to mitigate information asymmetry as information available to capital market participants increases. In this context, this study examines the effect of financial analysts’ cash flow forecasting on information asymmetry. According to previous studies, incentives to provide cash flow forecasts differ according to the information environment. This study examine whether there is a difference in the relationship between financial analysts’ cash flow forecasting activities and information asymmetry according to the quality of earnings by setting the quality of earnings as a substitute for information environment. According to the results of analysis, information asymmetry decreased more when financial analysts provided both earnings forecasts and cash flow forecasts than when they provided only earnings forecasts. In addition, information asymmetry was relieved more as the frequency of financial analysts’ cash flow forecasting increased and the cash flow forecasting ratio was higher. These results mean that financial analysts provide useful information to capital market participants through cash flow forecasting activities leading to decreases in information asymmetry. In addition, this study indicates that in case where the quality of earnings was low, information asymmetry was relieved further as financial analysts provided more cash flow forecasts, the frequency of cash flow forecasting increased, and the cash flow forecasting ratio was higher. This means that when the quality of earnings is lower, the negative(-) relations between cash flow forecasting activities and information asymmetry are reinforced and that when the quality of earnings is lower, the incentives for financial analysts to provide cash flow forecasts are higher. This study should be contributive in that it provides empirical results indicating that financial analysts’ cash flow forecasting activities contribute to the relief of information asymmetry. In addition, this study is meaningful in that it identified that financial analysts are providing cash flow forecasts to capital market participants with a view to delivering useful information as informediaries. In addition this study is differentiated from previous studies in that, whereas previous focused on financial analysts’ earnings forecasts, this study focuses on financial analysts’ cash flow forecasts. This study should provide additional evidence for the incentives for financial analysts’ provision of cash flow forecasts. 본 연구의 목적은 재무분석가가 현금흐름예측 활동을 많이 수행할수록 정보비대칭이 감소되는지를 살펴보는 것이다. 구체적으로 본 연구는 재무분석가의 현금흐름예측치 제공여부, 현금흐름예측 빈도(보고서 수) 및 이익예측 대비 현금흐름예측 빈도비율(이하 ‘현금흐름예측 비율’)에 따라 정보비대칭이 완화되는지를 분석한다. 아울러 이익의 질에 따라 현금흐름예측 활동과 정보비대칭 간의 관련성에 차이가 존재하는지를 살펴본다. 재무분석가는 우월한 능력과 전문지식을 바탕으로 자본시장참여자들에게 유용한 예측정보를 제공함으로써 자본시장의 정보비대칭을 완화시키는 역할을 수행한다. 재무분석가가 현금흐름예측치를 제공할 경우 현금흐름예측치 등 다양한 정보가 자본시장참여자들에게 제공된다. 즉, 재무분석가의 현금흐름예측치 제공으로 인해 자본시장참여자들이 이용가능한 정보가 증가하므로 정보비대칭이 완화될 것으로 기대된다. 이러한 배경에서 본 연구는 재무분석가의 현금흐름예측 확동이 정보비대칭에 미치는 영향을 살펴보고자 한다. 또한 선행연구에 따르면 정보환경에 따라 현금흐름예측치 제공유인이 다르게 나타난다. 이에 따라 본 연구는 정보환경의 대용치로 이익의 질을 설정하여 이익의 질에 따라 현금흐름예측 활동과 정보비대칭 간의 관련성에 차이가 있는지를 추가적으로 살펴본다. 분석결과에 따르면 재무분석가가 이익예측치와 현금흐름예측치를 함께 제공할 경우에 이익예측치만 제공하는 경우보다 정보비대칭이 감소하는 것으로 나타났다. 그리고 재무분석가의 현금흐름예측 빈도가 증가할수록, 현금흐름예측 비율이 높을수록 정보비대칭이 완화된다는 결과를 확인하였다. 이러한 결과는 재무분석가가 현금흐름예측 활동을 통해 자본시장참여자들에게 유용한 정보를 전달하며, 이로 인해 정보비대칭이 감소된다는 것을 의미한다. 또한 본 연구는 이익의 질이 낮은 경우 재무분석가가 현금흐름예측치를 제공할수록, 현금흐름예측 빈도가 증가할수록, 현금흐름예측 비율이 높을수록 정보비대칭이 더 크게 완화된다는 결과를 보여준다. 이는 이익의 질이 낮을수록 현금흐름예측 활동과 정보비대칭 간 음(-)의 관련성이 강화된다는 것을 의미하며, 이익의 질이 낮은 경우 재무분석가가 현금흐름예측치를 제공할 유인이 높다는 것을 나타낸다. 본 연구는 재무분석가의 현금흐름예측 활동이 정보비대칭 완화에 기여한다는 실증적 결과를 제공한다는 점에서 공헌점을 가질 것이다. 더불어 재무분석가가 정보중개인으로써 자본시장참여자들에게 유용한 정보를 전달하기 위해 현금흐름예측치를 제공하고 있음을 확인하였다는 점에서 의의가 있다. 또한 선행연구들이 재무분석가의 이익예측치에 초점을 두고 있는 반면, 본 연구는 재무분석가의 현금흐름예측치에 초점을 둔다는 점에서 차별성을 갖는다. 본 연구는 재무분석가의 현금흐름예측치 제공유인에 대한 추가적인 증거를 제시할 것이다.

수요 예측 평가를 위한 가중절대누적오차지표의 개발

최대일(Dea-Il Choi)옥창수(Chang-Soo Ok) 한국산업경영시스템학회 2015 한국산업경영시스템학회지 Vol.38 No.3

Aggregate Production Planning determines levels of production, human resources, inventory to maximize company’s profits and fulfill customer's demands based on demand forecasts. Since performance of aggregate production planning heavily depends on accuracy of given forecasting demands, choosing an accurate forecasting method should be antecedent for achieving a good aggregate production planning. Generally, typical forecasting error metrics such as MSE (Mean Squared Error), MAD (Mean Absolute Deviation), MAPE (Mean Absolute Percentage Error), and CFE (Cumulated Forecast Error) are utilized to choose a proper forecasting method for an aggregate production planning. However, these metrics are designed only to measure a difference between real and forecast demands and they are not able to consider any results such as increasing cost or decreasing profit caused by forecasting error. Consequently, the traditional metrics fail to give enough explanation to select a good forecasting method in aggregate production planning. To overcome this limitation of typical metrics for forecasting method this study suggests a new metric, WACFE (Weighted Absolute and Cumulative Forecast Error), to evaluate forecasting methods. Basically, the WACFE is designed to consider not only forecasting errors but also costs which the errors might cause in for Aggregate Production Planning. The WACFE is a product sum of cumulative forecasting error and weight factors for backorder and inventory costs. We demonstrate the effectiveness of the proposed metric by conducting intensive experiments with demand data sets from M3-competition. Finally, we showed that the WACFE provides a higher correlation with the total cost than other metrics and, consequently, is a better performance in selection of forecasting methods for aggregate production planning.

Evaluation of forecasting methods in aggregate production planning: A Cumulative Absolute Forecast Error (CAFE)

Ha, Chunghun,Seok, Hyesung,Ok, Changsoo Elsevier 2018 COMPUTERS & INDUSTRIAL ENGINEERING Vol.118 No.-

<P><B>Abstract</B></P> <P>The purpose of forecasting error measures is to estimate forecasting methods and choose the best one. Most typical forecasting error measures are designed based on the gap between forecasts and actual demands and, consequently, a forecasting method yielding forecasts in accordance with real demands is considered as good. However, in some applications such as aggregate production planning, these measures are not suitable because they are not capable for considering any effects caused by forecasting error such as increasing cost or decreasing profit. To tackle this issue, we propose a new measure, CAFE (Cumulative Absolute Forecast Error), to evaluate forecasting methods in terms of total cost. Basically, the CAFE is designed to consider not only forecasting errors but also costs occured by errors in aggregate production planning which is set up based on forecasts. The CAFE is a product sum of cumulative forecasting error and weight factors for backorder and inventory costs. We have demonstrated the effectiveness of the proposed measure by conducting intensive experiments with demand data sets from M3-competition.</P> <P><B>Highlights</B></P> <P> <UL> <LI> A new measure for forecasting error is designed to consider the cost which the error causes. </LI> <LI> The proposed measure enables to calculate the real effect of forecasting error. </LI> <LI> Extensive experiments with datasets from M3 competition. </LI> </UL> </P>

사례기반 추론기법과 인공신경망을 이용한 서비스 수요예측 프레임워크

황유섭(Yousub Hwang) 한국지능정보시스템학회 2012 지능정보연구 Vol.18 No.4

To enhance the competitive advantage in a constantly changing business environment, an enterprise management must make the right decision in many business activities based on both internal and external information. Thus, providing accurate information plays a prominent role in management’s decision making. Intuitively, historical data can provide a feasible estimate through the forecasting models. Therefore, if the service department can estimate the service quantity for the next period, the service department can then effectively control the inventory of service related resources such as human, parts, and other facilities. In addition, the production department can make load map for improving its product quality. Therefore, obtaining an accurate service forecast most likely appears to be critical to manufacturing companies. Numerous investigations addressing this problem have generally employed statistical methods, such as regression or autoregressive and moving average simulation. However, these methods are only efficient for data with are seasonal or cyclical. If the data are influenced by the special characteristics of product, they are not feasible. In our research, we propose a forecasting framework that predicts service demand of manufacturing organization by combining Case-based reasoning (CBR) and leveraging an unsupervised artificial neural network based clustering analysis (i.e., Self-Organizing Maps; SOM). We believe that this is one of the first attempts at applying unsupervised artificial neural network-based machine-learning techniques in the service forecasting domain. Our proposed approach has several appealing features : (1) We applied CBR and SOM in a new forecasting domain such as service demand forecasting. (2) We proposed our combined approach between CBR and SOM in order to overcome limitations of traditional statistical forecasting methods and We have developed a service forecasting tool based on the proposed approach using an unsupervised artificial neural network and Case-based reasoning. In this research, we conducted an empirical study on a real digital TV manufacturer (i.e., Company A). In addition, we have empirically evaluated the proposed approach and tool using real sales and service related data from digital TV manufacturer. In our empirical experiments, we intend to explore the performance of our proposed service forecasting framework when compared to the performances predicted by other two service forecasting methods; one is traditional CBR based forecasting model and the other is the existing service forecasting model used by Company A. We ran each service forecasting 144 times; each time, input data were randomly sampled for each service forecasting framework. To evaluate accuracy of forecasting results, we used Mean Absolute Percentage Error (MAPE) as primary performance measure in our experiments. We conducted one-way ANOVA test with the 144 measurements of MAPE for three different service forecasting approaches. For example, the F-ratio of MAPE for three different service forecasting approaches is 67.25 and the p-value is 0.000. This means that the difference between the MAPE of the three different service forecasting approaches is significant at the level of 0.000. Since there is a significant difference among the different service forecasting approaches, we conducted Tukey’s HSD post hoc test to determine exactly which means of MAPE are significantly different from which other ones. In terms of MAPE, Tukey’s HSD post hoc test grouped the three different service forecasting approaches into three different subsets in the following order: our proposed approach > traditional CBR-based service forecasting approach > the existing forecasting approach used by Company A. Consequently, our empirical experiments show that our proposed approach outperformed the traditional CBR based forecasting model and the existing service forecasting model used by Company A. T

데이터 예측을 위한 통계적 방법 비교 및 활용

신현철 건강보험심사평가원 심사평가정책연구소 2021 연구보고서 Vol.2021 No.0

Healthcare data forecasting is used in a variety of fields, including policy evaluation, health insurance financial estimation, and the detection of unusual claims symptoms. Furthermore, the evidence generated by the predictive model is critical in the development of health policies and decision-making processes. As a result, the prediction process must be scientific and systematic, and accuracy measures should be a key selection criterion for forecasting methods. By examining the data prediction methods used in the health care field as well as the most recent forecasting techniques, we attempted to suggest a forecasting method suitable for the characteristics of healthcare data in this study. To that end, a literature review, investigations on the most recent forecasting methodologies, and a comparative analysis of forecasting performance based on data type were performed. And the results of the analysis were synthesized to suggest a forecasting method appropriate for the type of healthcare data. The subject of forecasting method review included regression models, time series models, and machine learning. The data types were separated into continuous variables and count variables, and data sets of 12, 24, and 30 sizes were created. Health insurance claim data was used to compare forecasting performance, and SAS and Python were used as analysis tools. According to the findings, the machine learning forecasting method performed best for both continuous and count data types, while the ARIMA, time series analysis method performed reasonably well for continuous variables. Forecasting performance improved as the number of data points increased. In this study, we recommend a sample size of at least 30 subjects. This research is anticipated to help in the selection of an appropriate forecasting method for performing complex prediction tasks.

간헐적 수요예측을 위한 이항가중 지수평활 방법

하정훈 한국산업경영시스템학회 2018 한국산업경영시스템학회지 Vol.41 No.1

Intermittent demand is a demand with a pattern in which zero demands occur frequently and non-zero demands occur sporadically. This type of demand mainly appears in spare parts with very low demand. Croston’s method, which is an initiative intermittent demand forecasting method, estimates the average demand by separately estimating the size of non-zero demands and the interval between non-zero demands. Such smoothing type of forecasting methods can be suitable for mid-term or long-term demand forecasting because those provides the same demand forecasts during the forecasting horizon. However, the smoothing type of forecasting methods aims at short-term forecasting, so the estimated average forecast is a factor to decrease accuracy. In this paper, we propose a forecasting method to improve short-term accuracy by improving Croston’s method for intermittent demand forecasting. The proposed forecasting method estimates both the non-zero demand size and the zero demands’ interval separately, as in Croston’s method, but the forecast at a future period adjusted by binomial weight according to occurrence probability. This serves to improve the accuracy of short-term forecasts. In this paper, we first prove the unbiasedness of the proposed method as an important attribute in forecasting. The performance of the proposed method is compared with those of five existing forecasting methods via eight evaluation criteria. The simulation results show that the proposed forecasting method is superior to other methods in terms of all evaluation criteria in short-term forecasting regardless of average size and dispersion parameter of demands. However, the larger the average demand size and dispersion are, that is, the closer to continuous demand, the less the performance gap with other forecasting methods.