Attribute interaction aware matrix factorization method for recommendation

Matrix factorization (MF) models are effective and easy to expand and are widely used in industry, such as rating prediction and item recommendation. The basic MF model is relatively simple. In practical applications, side information such as attributes or implicit feedback is often combined to improve accuracy by modifying the model and optimizing the algorithm. In this paper, we propose an attribute interaction-aware matrix factorization (AIMF) method for recommendation tasks. We partition the original rating matrix into different sub-matrices according to the attribute interactions, train each sub-matrix independently, and merge all the latent vectors to generate the final score. Since the generated sub-matrices vary in size, an adaptive regularization coefficient optimization strategy and an adaptive latent vector dimension optimization strategy are proposed for sub-matrix training, and a variety of latent vector merging methods are put forward. The method AIMF has two advantages. When the original rating matrix is particularly large, the training time complexity of the MF-based model becomes higher and the update cost of the model is also higher. In AIMF, because each sub-matrix is usually much smaller than the original rating matrix, the training time complexity is greatly reduced after using parallel computing technology. Secondly, in AIMF, it is not necessary to modify the matrix factorization model to incorporate attributes and their interactive information into the model to improve the performance. The experimental results on the two classic public datasets MovieLens 1M and MovieLens 100k show that AIMF can not only effectively improve the accuracy of recommendation, but also make full use of parallel computing technology to improve training efficiency without modifying the matrix factorization model.

A methodology to develop a regional franchising strategy

Subject. This article discusses the strategy-related patterns concerning the development of the franchising system as an effective tool for business development. Objectives. The article aims to organize approaches to the evaluation of regional brands and trademarks promising for development under the franchise model, and develop a regional franchising strategy. Methods. For the study, I used analysis and synthesis, logical and systems approaches, and the theoretical generalization method. Results. The article offers an original rating of the subjects of the Russian Federation promising for the development and implementation of franchise offers, and a pattern of the company's development strategy based on the model of franchising. Conclusions. For the franchise network to be successful, it is necessary to carefully develop a strategy for regional franchising development.

The Consistency of Composite Ratings of Teacher Effectiveness: Evidence From New Mexico

Contemporary teacher evaluation systems use multiple measures of performance to construct ratings of teacher quality. While the properties of constituent measures have been studied, little is known about whether composite ratings themselves are sufficiently reliable to support high-stakes decision making. We address this gap by estimating the consistency of composite ratings of teacher quality from New Mexico’s teacher evaluation system from 2015 to 2016. We estimate that roughly 40% of teachers would receive a different composite rating if reevaluated in the same year; 97% of teachers would receive ratings within ±1 level of their original rating. We discuss mechanisms by which policymakers can improve rating consistency, and the implications of those changes to other properties of teacher evaluation systems.

Collaborative Filtering Recommendation Algorithm Integrated into Co-Rating Impact Factor

The sparsity rating data is one of the main challenges of recommendation system. For this problem, we presented a collaborative filtering recommendation algorithm integrated into co-ratings impact factor. The method reduced the sparsity of rating matrix by filling the original rating matrix. It made the full use of rating information and took the impact on similarity of co-ratings between users into consideration when looking for the nearest neighbor so that the similarities were accurately computed. Experimental results showed that the proposed algorithm, to some extent, improved the recommendation accuracy.

FAKTOR PENENTU PERINGKAT SUKUK

With the development of sukuk market as the Islamic alternatives of the existing bond market, the issue of how to assign a rating to the sukuk issuance rises. This study tries to provide an empirical foundation for the investors to estimate the ratings assign. Using approach from several rating agencies, past researches on bond ratings, financial distress prediction and bankruptcy prediction models, this study is trying to innovate a new model on determining the sukuk ratings. It used Multinomial Logit regression to create a model of rating probability from several theoretical variables, ie. firm size, leverage, profitability, fixed payment coverage, reputation and existence of guarantor. The result shows 80% of all valid cases are correctly classified into their original rating classes.Keywords: Sukuk, rating.JEL Classification: C35, E43, P43

DETERMINANT OF SUKUK RATINGS

With the development of sukuk market as the Islamic alternatives of the existing bond market, the issue of how to assign a rating to the sukuk issuance rises. This study tries to provide an empirical foundation for the investors to estimate the ratings assign. Using approach from several rating agencies, past researches on bond ratings, financial distress prediction and bankruptcy prediction models, this study is trying to innovate a new model on determining the sukuk ratings. It used Multinomial Logit regression to create a model of rating probability from several theoretical variables, ie. firm size, leverage, profitability, fixed payment coverage, reputation and existence of guarantor. The result shows 80% of all valid cases are correctly classified into their original rating classes.JEL Classification: C35, E43, P43Keywords: Sukuk, rating.

Erroneous Impairment Ratings

A nationwide study in 2005 of 2100 cases referred for impairment rating review found 80% to be erroneous, and 89% of these erroneous ratings were higher than appropriate. Among whole person erroneous ratings (839 of 1037 cases critiqued), the original physician's rating averaged 15.5% whole person permanent impairment, but following rerating by physician experts, the corrected rating averaged 5.6%; only 7% of the cases were underrated. All ratings were based on the AMA Guides to the Evaluation of Permanent Impairment, (AMA Guides), Fifth Edition. Tables and figures show error rates according to portion of the body affected, expert vs original rating, and other explanatory variables. Two physicians who use the AMA Guides should arrive at similar conclusions about impairment ratings, but most physicians have not received instruction about assessing impairment, disability, or causation and lack an adequate ability to assess these issues. Causation requires a given cause and a given effect that are associated with a reasonable degree of medical probability and also requires documentation with appropriate scientific evidence (not self-reports or historical time frames). Those who prepare and review assessments of impairment should ensure that clinical causation assessments were accurate, that the rating was performed at maximal medical improvement, that examination findings were consistent, and that the individual's normal state was determined.

Legal Update: Catastrophic Impairment: Catastrophic Impairment: Use of the AMA Guides in Ontario

Chapter 18, Pain, is the first attempt in the AMA Guides to the Evaluation of Permanent Impairment (AMA Guides), Fifth Edition, to define comprehensive pain evaluation and rating. Chapter 18 focuses on situations in which the pain itself is a major cause of suffering, dysfunction, or medical intervention even though an examining physician can find no demonstrable active disease or autonomic changes. These statements likely would be rejected by rating systems that require a physiologic basis for rating. Other chapters in the AMA Guides typically account for pain as part of specific impairment ratings; for example, a physician who rates cervical spine impairment must decide if a rating for pain beyond the original rating is appropriate. Many rating systems would reject such a rating for pain because of its subjective nature, but, despite these difficulties, even restrictive rating systems may require the use of Chapter 18 because, according to the AMA Guides, no other available method exists for considering impairment of certain conditions such as postherpetic neuralgias or migraines. As the AMA Guides acknowledges, Chapter 18 seems to allow impairment ratings caused by pain and suffering but at the same time to forestall actually assigning a number. One hopes for refinements in our understanding of chronic pain treatment and outcomes; in the interim, knowledgeable raters should guide nonmedical personnel toward a reasonable understanding of this chapter.

USING DATA TO ENHANCE THE EXPERT PANEL PROCESS

Objective: To enhance the validity of a well-known expert panel process, we used data from patient surveys to identify and correct rating errors.Methods: We used the two-round RAND/UCLA panel method to rate indications of harmful (presence of problems), hazardous (at risk for problems), and nonhazardous (no known risks) drinking in older adults. Results from the panel provided guidelines for classifying older individuals as harmful, hazardous, or nonhazardous drinkers, using a survey. The classifications yielded unexpectedly high numbers of harmful and hazardous drinkers. We hypothesized possible misclassifications of drinking risks and used the survey data to identify indications that may have led to invalid ratings. We modified problematic indications and asked three clinician panelists to evaluate the clinical usefulness of the modifications in a third panel round. We revised the indications based on panelist response and reexamined drinking classifications.Results: Using the original indications, 48% of drinkers in the sample were classified as harmful, 31% as hazardous, and 21% as nonhazardous. A review of the indications revealed framing bias in the original rating task and vague definitions of certain symptoms and conditions. The modified indications resulted in classifications of 22% harmful, 47% hazardous, and 31% nonhazardous drinkers.Conclusions: Analysis of survey data led to identification and correction of specific errors occurring during the panel-rating process. The validity of the RAND/UCLA method can be enhanced using data-driven modifications.