Purchase Sequences Research Articles

Fusing temporal and semantic dependencies for session-based recommendation

Session-based recommendation (SBR) predicts the next item in user sequences. Existing research focuses on item transition patterns, neglecting semantic information dependencies crucial for understanding users’ preferences. Incorporating semantic characteristics is vital for accurate recommendations, especially in applications like user purchase sequences. In this paper, to tackle the above issue, we novelly propose a framework that hierarchically fuses temporal and semantic dependencies. Technically, we present the Item Transition Dependency Module and Semantic Dependency Module based on the whole session set: (i) Item Transition Dependency Module is exclusively to learn the item embeddings through temporal relations and utilizes item transitions from both global and local levels; (ii) Semantic Dependency Module develops mutually independent embeddings of both sessions and items via stable interaction relations. In addition, under the unified organization of the Cross View, semantic information is adaptively incorporated into the temporal dependency learning and used to improve the performance of SBR. Extensive experiments on three large-scale real-world datasets show the superiority of our framework over current state-of-the-art methods. In particular, our model improves its performance over SOTA on all three datasets, with 5.5%, 0.2%, and 3.0% improvements on Recall@20, and 5.8%, 4.6%, and 2.0% improvements on MRR@20, respectively.

Hierarchical representation learning for next basket recommendation

The task of next basket recommendation is pivotal for recommender systems. It involves predicting user actions, such as the next product purchase or movie selection, by exploring sequential purchase behavior and integrating users’ general preferences. These elements may converge and influence users’ subsequent choices. The challenge intensifies with the presence of varied user purchase sequences in the training set, as indiscriminate incorporation of these sequences can introduce superfluous noise. In response to these challenges, we propose an innovative approach: the Selective Hierarchical Representation Model (SHRM). This model effectively integrates transactional data and user profiles to discern both sequential purchase transactions and general user preferences. The SHRM’s adaptability, particularly in employing nonlinear aggregation operations on user representations, enables it to model complex interactions among various influencing factors. Notably, the SHRM employs a Recurrent Neural Network (RNN) to capture extended dependencies in recent purchasing activities. Moreover, it incorporates an innovative sequence similarity task, grounded in a k-plet sampling strategy. This strategy clusters similar sequences, significantly mitigating the learning process’s noise impact. Through empirical validation on three diverse real-world datasets, we demonstrate that our model consistently surpasses leading benchmarks across various evaluation metrics, establishing a new standard in next-basket recommendation.

Marketing Decision Model and Consumer Behavior Prediction With Deep Learning

This article presents a study using ResNet-50, GRU, and transfer learning to construct a marketing decision-making model and predict consumer behavior. Deep learning algorithms address the scale and complexity of consumer data in the information age. Traditional methods may not capture patterns effectively, while deep learning excels at extracting features from large datasets. The research aims to leverage deep learning to build a marketing decision-making model and predict consumer behavior. ResNet-50 analyzes consumer data, extracting visual features for marketing decisions. GRU model temporal dynamics, capturing elements like purchase sequences. Transfer learning improves performance with limited data by using pre-trained models. By comparing the model predictions with ground truth data, the performance of the models can be assessed and their effectiveness in capturing consumer behavior and making accurate predictions can be measured. This research contributes to marketing decision-making. Deep learning helps understand consumer behavior, formulate personalized strategies, and improve promotion and sales. It introduces new approaches to academic marketing research, fostering collaboration between academia and industry.

Customer segmentation by web content mining

This article introduces a new dimension, Interpurchase Time (T), into the existing RFM (Recency, Frequency, and Monetary) model to form an expanded RFMT model for parsing consumers' online purchase sequences in a long period to implement customer segmentation. The proposed RFMT model can track and discern changes in customer purchasing behaviors during their whole shopping cycle. Firstly, a web content retrieving system was developed to fetch publicly available customer data on a retailer's website, including demographic information (gender, age, location, etc.) and product information (name, price, date, etc.) of each purchase in a period from 2008 to 2019. The RFMT values of a customer were then computed from the retrieved data and subsequently analyzed by the hierarchical clustering to derive seven homogeneous clusters with specific customer profiles. Subsequently, demographic features and product preferences were identified for each cluster with business insights that can help the retailer to improve customer relationships and to implement targeted recommendation strategies.

Dig users’ intentions via attention flow network for personalized recommendation

Accurately forecasting user’s purchase intention over time is a huge challenge for personalized recommend systems, of which a critical problem is how to model the changes of user preference and temporal correlation of items. In this paper, aiming at addressing this question, we first introduce attention flow network to model users’ purchase records by leveraging attention flow that describes the changing process of purchase intention. Then based on the attention flow network and individuals’ attention flows, we propose a novel personalized recommendation algorithm named Attention Flow Network based Personalized Recommendation (AFNPR). Our method integrates all the purchase sequences of users into a weighted attention flow network, and recommends items based on transition probabilities related to attention flow network with user’s attention decay, which is efficient in linear time. The experiments demonstrate its superior performance on several real datasets.

Sequences of purchases in credit card data reveal lifestyles in urban populations

Zipf-like distributions characterize a wide set of phenomena in physics, biology, economics, and social sciences. In human activities, Zipf's law describes, for example, the frequency of appearance of words in a text or the purchase types in shopping patterns. In the latter, the uneven distribution of transaction types is bound with the temporal sequences of purchases of individual choices. In this work, we define a framework using a text compression technique on the sequences of credit card purchases to detect ubiquitous patterns of collective behavior. Clustering the consumers by their similarity in purchase sequences, we detect five consumer groups. Remarkably, post checking, individuals in each group are also similar in their age, total expenditure, gender, and the diversity of their social and mobility networks extracted from their mobile phone records. By properly deconstructing transaction data with Zipf-like distributions, this method uncovers sets of significant sequences that reveal insights on collective human behavior.

A grocery recommendation for off-line shoppers

PurposeMany off-line retailers have experienced a slump in sales and have the potential risk of overstock or understock. To overcome these problems, retailers have applied data mining techniques, such as association rule mining or sequential association rule mining, to increase sales and predict product demand. However, because these techniques cannot generate shopper-centric rules, many off-line shoppers are often inconvenienced after writing their shopping lists carefully and comprehensively. Therefore, the purpose of this paper is to propose a personalized recommendation methodology for off-line grocery shoppers.Design/methodology/approachThis paper employs a Markov chain model to generate recommendations for the shopper’s next shopping basket. The proposed methodology is based on the knowledge of both purchased products and purchase sequences. This paper compares the proposed methodology with a traditional collaborative filtering (CF)-based system, a bestseller-based system and a Markov-chain-based system as benchmark systems.FindingsThe proposed methodology achieves improvements of 15.87, 14.06 and 37.74 percent with respect to the CF-, Markov chain-, and best-seller-based benchmark systems, respectively, meaning that not only the purchased products but also the purchase sequences are important elements in the personalization of grocery recommendations.Originality/valueMost of the previous studies on this topic have proposed on-line recommendation methodologies. However, because off-line stores collect transaction data from point-of-sale devices, this research proposes a methodology based on purchased products and purchase patterns for off-line grocery recommendations. In practice, this study implies that both purchased products and purchase sequences are viable elements in off-line grocery recommendations.

Discriminating between behaviour using market data from panels

Considerable research exists on stochastic models of switching behaviour that uses sequences of individual-level purchase data. While at the individual level, sample size and sequence length are limiting factors, at the aggregate level, heterogeneity with respect to purchase sequences may assist in interpreting results. The authors propose an approach to discriminate between the switching behaviour of variety seeking, indifference and reinforcement. Only the proportion of 100% loyal customers, market share data and an estimation of the promotional effect -information all available from consumer panel data - are necessary to fit the model.

Predicting partial customer churn using Markov for discrimination for modeling first purchase sequences

Currently, in order to remain competitive companies are adopting customer centered strategies and consequently customer relationship management is gaining increasing importance. In this context, customer retention deserves particular attention. This paper proposes a model for partial churn detection in the retail grocery sector that includes as a predictor the similarity of the products’ first purchase sequence with churner and non-churner sequences. The sequence of first purchase events is modeled using Markov for discrimination. Two classification techniques are used in the empirical study: logistic regression and random forests. A real sample of approximately 95,000 new customers is analyzed taken from the data warehouse of a European retailing company. The empirical results reveal the relevance of the inclusion of a products’ sequence likelihood in partial churn prediction models, as well as the supremacy of logistic regression when compared with random forests.

Sequential market basket analysis

Market basket analysis (MBA) is a powerful and common practice in modern retailing that has some limitations stemming from the fact that it infers purchase sequence from joint-purchasing data. However, internet retailers automatically collect purchase-sequence data from their shoppers, and new technology is available for traditional (bricks and mortar) retailers to do the same, making it possible to analyze purchase sequences, rather than inferring them from joint purchases. This study first compares and contrasts traditional market basket analysis with a sequential extension, and then proposes a framework for purchase-sequence analysis, which is illustrated utilizing shopping trip data from one grocery store.

Modeling partial customer churn: On the value of first product-category purchase sequences

Retaining customers has been considered one of the most critical challenges among those included in Customer Relationship Management (CRM), particularly in the grocery retail sector. In this context, an accurate prediction whether or not a customer will leave the company, i.e. churn prediction, is crucial for companies to conduct effective retention campaigns. This paper proposes to include in partial churn detection models the succession of first products’ categories purchased as a proxy of the state of trust and demand maturity of a customer towards a company in grocery retailing. Motivated by the importance of the first impressions and risks experienced recently on the current state of the relationship, we model the first purchase succession in chronological order as well as in reverse order, respectively. Due to the variable relevance of the first customer–company interactions and of the most recent interactions, these two variables are modeled by considering a variable length of the sequence. In this study we use logistic regression as the classification technique. A real sample of approximately 75,000 new customers taken from the data warehouse of a European retail company is used to test the proposed models. The area under the receiver operating characteristic curve and 1%, 5% and 10% percentiles lift are used to assess the performance of the partial-churn prediction models. The empirical results reveal that both proposed models outperform the standard RFM model.

Housing Purchases and the Dynamics of Housing Wealth

I model the purchase behavior of main and secondary housing by Spanish households using the panel sample from the first two waves of the Spanish household finance survey (EFF). I estimate discrete hazard models using retrospective and within-period purchase sequences. I also estimate an (S,s) model combining transactions data with longitudinal information on household wealth and housing stock values. I look at the role of adaptive expectations about the rate of return on housing and find they have a positive and significant effect on the demand for houses. This is true for historical and within-period purchase probabilities as well as for the target ratio of housing wealth to total wealth. The volatility of house price growth has a negative effect on purchases for investment but a positive one on purchases for consumption.

A hybrid of sequential rules and collaborative filtering for product recommendation

Customers’ purchase behavior may vary over time. Traditional collaborative filtering (CF) methods make recommendations to a target customer based on the purchase behavior of customers whose preferences are similar to those of the target customer; however, the methods do not consider how the customers’ purchase behavior may vary over time. In contrast, the sequential rule-based recommendation method analyzes customers’ purchase behavior over time to extract sequential rules in the form: purchase behavior in previous periods ⇒ purchase behavior in the current period. If a target customer’s purchase behavior history is similar to the conditional part of the rule, then his/her purchase behavior in the current period is deemed to be the consequent part of the rule. Although the sequential rule method considers the sequence of customers’ purchase behavior over time, it does not utilize the target customer’s purchase data for the current period. To resolve the above problems, this work proposes a novel hybrid recommendation method that combines the segmentation-based sequential rule method with the segmentation-based KNN-CF method. The proposed method uses customers’ RFM (Recency, Frequency, and Monetary) values to cluster customers into groups with similar RFM values. For each group of customers, sequential rules are extracted from the purchase sequences of that group to make recommendations. Meanwhile, the segmentation-based KNN-CF method provides recommendations based on the target customer’s purchase data for the current period. Then, the results of the two methods are combined to make final recommendations. Experiment results show that the hybrid method outperforms traditional CF methods.

Investigating purchasing-sequence patterns for financial services using Markov, MTD and MTDg models

In the past, several authors have found evidence for the existence of a priority pattern of acquisition for durable goods, as well as for financial services. Its usefulness lies in the fact that if the position of a particular customer in this acquisition sequence is known, one can predict what service will be acquired next by that customer. In this paper, we analyse purchase sequences of financial services to identify cross-buying patterns, which might be used to discover cross-selling opportunities as part of customer relationship management (CRM). Hereby, special attention is paid to transitions, which might encourage bank-only or insurance-only customers to become financial-services customers. We introduce the mixture transition distribution (MTD) model as a parsimonious alternative to the Markov model for use in the analysis of marketing problems. An interesting extension on the MTD model is the MTDg model, which is able to represent situations where the relationship between each lag and the current state differs. We illustrate the MTD and MTDg model on acquisition sequences of customers of a major financial-services company and compare the fit of these models with that of the corresponding Markov model. Our results are in favor of the MTD and MTDg models. Therefore, the MTD as well as the MTDg transition matrices are investigated to reveal cross-buying patterns. The results are valuable to product managers as they clarify the customer flows among product groups. In some cases, the lag-specific transition matrices of the MTDg model give better insight into the acquisition patterns than the general transition matrix of the MTD model.

Mining changes in customer buying behavior for collaborative recommendations

The preferences of customers change over time. However, existing collaborative filtering (CF) systems are static, since they only incorporate information regarding whether a customer buys a product during a certain period and do not make use of the purchase sequences of customers. Therefore, the quality of the recommendations of the typical CF could be improved through the use of information on such sequences. In this study, we propose a new methodology for enhancing the quality of CF recommendation that uses customer purchase sequences. The proposed methodology is applied to a large department store in Korea and compared to existing CF techniques. Various experiments using real-world data demonstrate that the proposed methodology provides higher quality recommendations than do typical CF techniques, with better performance, especially with regard to heavy users.

An Investigation into the Order of the Brand Choice Process

This paper investigates the order of the brand choice process at the individual family level using a variety of tests: t, likelihood ratio, binomial runs, and multinomial runs tests. The test of stationarity developed permits separate analyses for stationary and nonstationary purchase sequences. Our analyses indicate that the purchase sequences of a majority of stationary consumers are consistent with the zero-order assumption. This result is observed for all the nine frequently-bought packaged goods studied with the use of panel data and for a product category (soft drinks) using experimental data.

A Probablistic Market Model of Purchase Timing and Brand Selection

A stochastic model of consumer purchase behavior for frequently purchased, low cost products is developed. Both brand selection and purchase timing are incorporated in the model; a first-order Markov process is used to describe brand selection, and Erlang density functions are used to describe time between purchases. The market's behavior is obtained by describing the individual consumer's behavior and then aggregating over consumers. The model's predictions of various aggregate purchase timing statistics and repeat purchase sequences are empirically verified.