Large Transaction Data Research Articles

Liquidity, Credit Risk, and Their Interaction on the Spreads in China’s Corporate Bond Market

This paper investigates the impact of multidimension liquidity, credit risk, and the interaction between liquidity and credit risk on corporate bond spreads based on a large transaction data set from July, 2006 to June, 2016, including the monthly data of 3716 bonds in China. Our main findings reveal that liquidity premiums are the main parts of corporate bond spreads. The interaction between liquidity and credit risk plays a significant role in determining corporate bond spreads. In addition, the differences between the interbank market and the exchange market have a significant impact on corporate bond spreads in normal period, and the interaction between liquidity and credit risk has an enhanced impact on corporate bond spreads during financial crisis. We also find that the interaction between liquidity and credit risk will increase with the increase of liquidity risk and credit risk and it is a time‐varying dynamic process.

IMPLEMENTASI DATA MINING UNTUK MENENTUKAN POLA PENJUALAN DENGAN MARKET BASKET ANALYSIS

The increasingly fierce competition in the business world lately has forced business actors to have the right strategy to increase their product sales, one of which is to determine sales patterns. However, along with the amount of sales transaction data that is carried out every day, it makes it difficult for business actors to analyze existing sales patterns. So we need a method that can help determine sales patterns from even large transaction data, one of which uses the Market Basket Analysis method which uses customer data that has been stored in the database to find new information in it. By utilizing the sales transaction data of consumers supported by the Market Basket Analysis method, it is possible to determine the right marketing tactics for business actors so as to increase sales

Extraction of Frequent Itemset Using Hybrid Algorithm for Large Transactional Data

Association rule mining is one of the significant strategies for bigdata to extract the knowledge from data. This Knowledge acts as a considerable part in the decision-making process. Especially for the business insights, So, this paper proposed a new Hybrid Ant Colony and African Buffalo Model which helps to discover the frequent item sets within less execution time and high accuracy. Moreover, Fitness function of this model is utilized to compute the support count of each item to attain the high accuracy rate. Finally, the strength of the association rules is measured using the key metrics support and confidence.

Effectiveness of frequent inventory audits in retail stores: an empirical evaluation

PurposeThe purpose of this paper is to focus on the effectiveness of the inventory audit process to manage operational issues related to inventory errors in retail stores. An evaluation framework is proposed based on developing an error profile of store inventory using product attributes and inventory information.Design/methodology/approachA store inventory error profile is developed using data on price, sales, popularity, replenishment cycle, inventory levels and inventory errors. A simulation model of store inventory management system grounded in empirical data is used to evaluate the effectiveness of the inventory audit process in a high SKU-variety retail store. The framework is tested using a large transaction data set comprised of over 200,000 records for 7,400 SKUs.FindingsThe results show that store inventory exhibits different inventory error profile groups that would determine the effectiveness of store inventory audits. The results also identify an interaction effect between store inventory policies and replenishment process that moderates the effectiveness of inventory audits.Research limitations/implicationsThe analysis is based on data collected from a single focal firm and does not cover all the different segments of the retail industry. However, the evaluation framework presented in the paper is fully generalizable to different retail settings offering opportunity for additional studies.Practical implicationsThe findings about the role of different error profile groups and the interaction effect of store audits with inventory and store replenishments would help retailers incorporate a more effective inventory audit process in their stores.Originality/valueThis paper presents a novel approach that uses store inventory profiles to evaluate the effectiveness of inventory audits. Unlike previous papers, it is the first empirical study in this area that is based on inventory error data gathered from multiple audits that identify the interaction effect of inventory policy and replenishments on the inventory audit process.

Application of Machine Learning and Grocery Transaction Data to Forecast Effectiveness of Beverage Taxation.

Sugar Sweetened Beverages (SSB) are the primary source of artificially added sugar and have a casual association with chronic diseases. Taxation of SSB has been proposed, but limited evidence exists to guide this public health policy. Grocery transaction data, with price, discounting and other information for beverage products, present an opportunity to evaluate the likely effects of taxation policy. Sales are often non-linearly associated with price and are affected by the prices of multiple competing brands. We evaluated the predictive performance of Boosted Decision Tree Regression (B-DTR) and Deep Neural Networks (DNN) that account for the non-linearity and competition across brands, and compared their performance to a benchmark regression, the Least Absolute Shrinkage and Selection Operator (LASSO). B-DTR and DNN showed a lower Mean Squared Error (MSE) of prediction in the sales of most major SSB brands in comparison to LASSO, indicating a superior accuracy in predicting the effectiveness of SSB taxation. We demonstrated the application of machine learning methods and large transactional data from grocery stores to forecast the effectiveness food taxation.

Champion-challenger analysis for credit card fraud detection: Hybrid ensemble and deep learning

Credit card fraud detection is an essential part of screening fraudulent transactions in advance of their authorization by card issuers. Although credit card frauds occur extremely infrequently, they result in huge losses as most fraudulent transactions have large values. An adequate detection of fraud allows investigators to take timely actions that can potentially prevent additional fraud or financial losses. In practice, however, investigators can only check a few alerts per day since the investigation process can be long and tedious. Thus, the primary goal of the fraud detection model is to return accurate alerts with fewer false alarms and missed frauds. Conventional fraud detection is mainly based on the hybrid ensemble of diverse machine learning models. Recently, several studies have compared deep learning and traditional machine learning models including ensemble. However, these studies used evaluation methods without considering that the real-world fraud detection system operated with the constraints: (i) the number of investigators who check the high-risk transactions from the data-driven scoring models are limited and (ii) the two types of misclassification, false alarms and missed frauds, have different costs. In this study, we conducted an in-depth comparison between the hybrid ensemble and deep learning method to determine whether or not to adopt the latter in our partner’s system that currently operates with the hybrid ensemble model. To compare the two, we introduced the champion-challenger framework and the development process of the two models. After developing the two models, we evaluated them on large transaction data sets taken from our partner, a major card issuing company in South Korea. We used various practical evaluation metrics appropriate for this domain that has severe class and cost imbalances. Moreover, we deployed these models in a real-world fraud detection system to check the post-launch performance for one month. The challenger outperformed the champion on both in off-line and post-launch tests.

Recommender system based on pairwise association rules

Recommender systems based on methods such as collaborative and content-based filtering rely on extensive user profiles and item descriptors as well as on an extensive history of user preferences. Such methods face a number of challenges; including the cold-start problem in systems characterized by irregular usage, privacy concerns, and contexts where the range of indicators representing user interests is limited. We describe a recommender algorithm that builds a model of collective preferences independently of personal user interests and does not require a complex system of ratings. The performance of the algorithm is analyzed on a large transactional data set generated by a real-world dietary intake recall system.

Impacts of Reward Accrual Effort on Redemption Behavior in a Multi-Vendor Loyalty Program

This research explores two key facets of behavior (reward point accrual and redemption) that consist of a loyalty program. It focuses on assessing the impact of accrual effort level on three types of redemption behavior: speed, unit size, and hedonic preference at the individual level by using large scale transaction data from a multi-vendor loyalty program providing flexible environment for point accrual and redemption. Findings from this research demonstrate that customers tend 1) to speed up point redemption, 2) to enlarge the size of redeemed points, and 3) to prefer utilitarian rewards as the level of effort at the accrual stage of reward point increases.

Construction and evaluation of structured association map for visual exploration of association rules

The association rule mining is one of the most popular data mining techniques, however, the users often experience difficulties in interpreting and exploiting the association rules extracted from large transaction data with high dimensionality. The primary reasons for such difficulties are two-folds. Firstly, too many association rules can be produced by the conventional association rule mining algorithms, and secondly, some association rules can be partly overlapped. This problem can be addressed if the user can select the relevant items to be used in association rule mining, however, there are often quite complex relations among the items in large transaction data. In this context, this paper aims to propose a novel visual exploration tool, structured association map (SAM), which enables the users to find the group of the relevant items in a visual way. The appearance of SAM is similar with the well-known cluster heat map, however, the items in SAM are sorted in more intelligent way so that the users can easily find the interesting area formed by a set of associated items, which are likely to constitute interesting many-to-many association rules. Moreover, this paper introduces an index called S2C, designed to evaluate the quality of SAM, and explains the SAM based association analysis procedure in a comprehensive manner. For illustration, this procedure is applied to a mass health examination result data set, and the experiment results demonstrate that SAM with high S2C value helps to reduce the complexities of association analysis significantly and it enables to focus on the specific region of the search space of association rule mining while avoiding the irrelevant association rules.

Impacts of Reward Accrual Effort on Redemption Speed, Unit Size and Hedonic Preference in a Multi-Vendor Loyalty Program

This research explores two key behaviors (reward point accrual and redemption) that consist of a loyalty program. The authors focus on assessing the impact of accrual effort level on three types of redemption behavior: speed, unit size, and hedonic preference at the individual level by using large scale transaction data from a multi-vendor loyalty program providing flexible point accrual and redemption opportunities. Our findings show that customers tend 1) to speed up point redemption, 2) to enlarge the size of redeemed points, and 3) to prefer utilitarian rewards as the level of effort during the accrual stage of reward point increases.

Models for association rules based on clustering and correlation

Association rules require models to understand their relationship to statistical properties of the data set. In this work, we study mathematical relationships between association rules and two fundamental techniques: clustering and correlation. Each cluster represents an important itemset. We show the sufficient statistics for clustering and correlation on binary data sets are the linear sum of points and the quadratic sum of points, respectively. We prove itemset support can be bounded and approximated from both models. Support bounds and support estimation obey the set downward closure property for fast bottom-up search for frequent itemsets. Both models can be efficiently computed with sparse matrix computations. Experiments with real and synthetic data sets evaluate model accuracy and speed. The clustering model is accurate to estimate support, given a sufficiently large number of clusters and it is more accurate than correlation, except for sets of two items. Accuracy increases as the number of clusters grows, but decreases as the minimum support threshold decreases. Once built, the clustering model represents a faster alternative than the traditional A-priori algorithm and the correlation model to mine associations. The correlation model is faster to compute than clustering, but it is less accurate. Time complexity to compute both models is linear on data set size, whereas dimensionality marginally impacts time when analyzing large transaction data sets.

Durations, volume and the prediction of financial returns in transaction time

Traditional microstructural theories of asset pricing emphasize the role of volume as a trend indicator. With the availability of large transaction data sets, one has started recently to incorporate more information of the trades, such as the time between trades, to describe the multivariate dynamics of transactions. Without knowing a priori the relation between the observed components of a trade—price, duration between trades, and volume—one may follow the principle of ‘letting the data speak for themselves’. The goal of this paper is to evaluate the informational content of both volume and durations to predict transaction returns using explorative non-parametric methods. The empirical results for transaction data of IBM stock prices confirm the role of volume as a trend indicator. After a sell (buy) expected returns are decreasing (increasing) with volume and increasing (decreasing) with durations. A.forecasting exercise shows that the superiority of the non-parametric model over simple parameterizations carries over to out-of-sample prediction.

Pricing size effects in housing markets

This paper examines whether pricing size effects as observed in securities markets exist in housing markets. A large transaction data set for the city of Perth, Western Australia is used to construct market aggregate and price quartile repeat-sales and hedonic indexes for the period 1988–96. Methodologies for the identification of pricing size effects are proposed and significant pricing size effects are observed. Cheaper properties exhibit higher rates of real price change in the short term but the lowest rates in the longer term. These results are consistent for tests with index models and individual property price changes. These results are shown to cause bias in transaction based indexes. A number of potential areas for future research are proposed.