The T (Taguchi) method is belongs to the MT (Mahalanobis Taguchi) system which is a representative method in quality engineering. The T method is used for prediction or estimation in fields such as economics and weather forecast.
Actual analyses often entail the presence of missing value, which make processing a model more complex. Overall, it is necessary to impute missing values with some other ones.
As highlighted by Masuda (2012), “Although not well-known, the T method has the merit that it is not in conventional multiple regression analysis. That is, even if missing data exists, it can be analyzed in the T method.” More specifically, the T method integrates the results of the single regression analysis, thus making it possible to perform analyses without processing missing values, if present. Then, based on this, the missing data analysis in the T method is taken into account.
In this study, we consider the development of the analysis of data including missing values. Specifically, we clarify the calculation procedure in the T method and its improvement with missing data, run simulations under various models, and compare the estimation accuracies.
In pattern recognition, there are two widely used methods, MT (Mahalanobis Taguchi) and machine learning. The advantage of the MT method is that it makes it possible to detect unknown abnormal patterns. However, the discrimination accuracy is relatively lower than machine learning. In addition, there is a constraint on the variables. Alternatively, machine learning has higher discrimination accuracy than the MT method; however, discrimination of unknown abnormal patterns may be difficult.
The purpose of this study is to propose a method that utilizes the strengths of both methods and eliminates the drawbacks of them. By applying AdaBoost to the MT method, the proposed method is evaluated in terms of the following. First, discrimination accuracy comparable to that obtained using machine learning is realized. Second, it is possible to accurately distinguish unknown abnormal patterns, which may pose a challenge when using machine learning. Third, the variable constraint condition in the MT method can be solved. Through analytics of breast cancer data and a created simulation data based on it, we demonstrate the superiority of the proposed method over conventional methods when the data is likely to include unknown abnormal patterns.
This paper proposes a method for analyzing geometrical characteristics, especially for improvement of manufacturing machining process. The definition of the geometrical characteristics in the GPS (Geometrical Product Specifications) standard defined as ISO 1101 are the value of difference from geometrically correct points, lines and planes as the datum, and it is aggregated property of multiple factors. This aggregated property of the geometrical characteristics is for conformity evaluation of the product functional requirement, and may not be appropriate to use for the improvement the manufacturing process. The geometrical characteristics have resolution structure and/or hierarchical structure. To elusive the causal mechanism and their relationship, and to use the causal mechanism for improvement of the manufacturing process, the geometrical characteristics should be resolved on their coordinate position, or should be observed the hierarchical structure. And as the first step, observation of the actual shape of the feature is also useful to analyze. Through the case study, this paper proposes the efficient procedure of process analysis of the geometrical characteristics.
Today internet shopping sites are widely used. On such sites, consumers cannot take products in their hands, so user reviews of products must be informative for the consumers. However, number of reviews are very large, so consumers cannot read them all. The may miss a very important information which can catch reading all reviews. Purpose of our research is to make a method of indicating only useful reviews for reducing consumer's burden.
We assume that useful reviews are different by consumers. We found that usefulness of reviews differs from consumers' purpose of browsing review, and also from consumers' knowledge about purchase target product group. We can classify consumers into six groups. Next we identified factors of useful reviews for each group. Finally we constructed models for evaluating usefulness of each reviews for each group.
We made dictionaries for each factor to judge which factors are included in each review.
For validating our models, we conducted questionnaires' survey. Consumers were shown two groups of user reviews, one group contains top 10 useful reviews based on our model, and the other contains original 10 reviews which shopping site shows. We could confirm that our model can show more useful reviews to consumers.
Japan is one of the most natural disaster-prone countries in the world. During a natural disaster or mass casualty incident, hospitals are likely to receive a large number of injured people. Some hospitals have established a Business Continuity Management System (BCMS) to increase healthcare resilience. One of the most important activities in a BCMS is an internal audit. This enables auditors to grasp certain problems effectively and efficiently by preparing audit items beforehand. Internal audits in BCMS target activities that take place during a disaster. There are some characteristics of disaster medicine, such as the restriction of physical and personnel resources, which do not exist in routine work. It is important to conduct an internal audit by considering these characteristics to improve the continuity of disaster medicine provision. This study aims to identify these characteristics and clarify the audit items for the BCMS. This study also demonstrates the results of an internal audit by applying the proposed audit items at an acute care hospital. Furthermore, although an internal audit has many possible focus areas, this study discusses audits from the perspective of disaster procedure manuals, which enable us to better grasp the details of disaster medicine.