http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
A Latent Class Analysis for Item Demand Based on Temperature Difference and Store Characteristics
Yuto Seko,Ryotaro Shimizu,Gendo Kumoi,Tomohiro Yoshikai,Masayuki Goto 대한산업공학회 2021 Industrial Engineeering & Management Systems Vol.20 No.1
In retail stores, there is an increasing need for predicting item demand using accumulated purchase history data to cope with the fluctuating consumer demands. These fluctuations in item demand are influenced by external factors and consumer preferences. Among these, store characteristics and weather conditions, which are closely related to consumer behavior, have strong effects on item demand. For this reason, it is very important to quantitatively grasp demand fluctuations of items that are influenced by changes in weather conditions for each store by using an integrated analysis of the purchase history data of many stores and weather conditions. In this research, we focus on the temperature difference, which is the average temperature difference from the previous day, as a weather condition affecting item sales. Because consumer feeling about a temperature is dependent on the temperature difference from the previous day, it is meaningful to construct a prediction model using this information. In this research, we propose a latent class model to express the relationship between weather conditions, store characteristics, and item demand fluctuation. Also, through an analysis experiment using an actual data set, we show the usefulness of the proposed model by extracting items that are influenced by weather conditions.
Tomoki Amano,Ryotaro Shimizu,Masayuki Goto 대한산업공학회 2023 Industrial Engineeering & Management Systems Vol.22 No.3
In recent years, recommender systems based on machine learning have become common tools on various web ser-vices. Among recommendation models, embedding methods such as Item2Vec that utilize embedded representations of items are widely used in actual applications due to their effectiveness and ease of use. By utilizing embedded rep-resentations acquired through learning the interaction between users and items, it is easy to discover similar items from the viewpoint of the user’s purchasing tendencies. In contrast, with this method, only biased items are recom-mended, making it difficult to ensure a wide variety of recommended items. However, there is a trade-off between the diversity of recommended items and accuracy and providing diversity in recommended items while maintaining accuracy is a challenging problem. Therefore, in this study, we propose a method to expand the new evaluation met-ric "recommendation region" (sum of distances of recommended items from the user vector in the embedding space) without significantly reducing accuracy. Specifically, we recommend not only items that are close to the user vector in the embedding space but also items with a certain distance based on detailed observation of the positional relation-ships. With the proposed method, we aim to increase user satisfaction by expanding the diversity of items that the user comes into contact with in the service. Finally, we demonstrate the usefulness of our proposed method through evaluation experiments using open-source datasets.
Optimizing FT-Transformer: Sparse Attention for Improved Performance and Interpretability
Tokimasa Isomura,Ryotaro Shimizu,Masayuki Goto 대한산업공학회 2024 Industrial Engineeering & Management Systems Vol.23 No.2
In recent studies, a class of deep learning models has been suggested to provide higher prediction accuracy for tabular data than gradient boosting algorithms, the current mainstream approach for such structured data. In particular, the effectiveness of FT-Transformer (FTT), which customizes the transformer model to tabular data, has been shown in recent research. Transformer was initially proposed for unstructured data and have shown high performance by sensitively considering the relationships between all features (e.g., words and patch images) through the attention mechanism. However, the relationships between input variables (features) and an output variable in tabular data can be assumed less complex than those in unstructured data. Therefore, we propose FTT+, an improved FTT suitable for tabular data that improves performance by not excessively considering the unnecessary relationship between features in the transformers attention mechanism. In addition, we expand FTT+ as a strong explainable model by taking a novel approach based on the transfer learning method. The effectiveness and interpretability of our proposals are clarified through evaluation experiments on regression, binary classification, multi-level classification tasks, and multiple analyses of these results. With this studys contribution, the proposed models could be suggested as the new standard for all tasks using tabular data.