Loading article...
Loading article...
Stream/Area:
-
Subject:
-
Paper ID:
IJSREM67757
Volume:
10 |
10 |
Oct 2026
Pages:
1-7
Publication Frequency:
Monthly (12 issues per year)
License:International Journal of Scientific Research in Engineering and Management follows the Creative Commons Attribution 4.0 (CC BY 4.0) License, allowing free use and distribution with proper attribution.
Masters of computer application
Yashaswi Education Socienty's International Institute of Management Science
Masters of computer application
Yashaswi Education Socienty's International Institute of Management Science
Masters of computer application
Yashaswi Education Socienty's International Institute of Management Science
Masters of computer application
Yashaswi Education Socienty's International Institute of Management Science
Retail companies use machine learning to sort products into sales categories that guide stocking and promotion decisions. Running such a classifier as a cloud service raises separate questions about latency, scaling and cost. This paper proposes SalesSense, an event-driven serverless design on AWS (S3, Lambda, API Gateway, SageMaker Serverless Inference, DynamoDB, SNS, CloudWatch, IAM) that connects product-sales classification to a decision-support workflow. The design has not been deployed. We report local experiments on three public retail datasets, with naive baselines and checks that remove product scale. On the two temporal datasets, a persistence rule that repeats last week’s class matched or beat every learned model: on Store Sales it scored 0.987 accuracy against 0.972 for the best model over three folds, and on Online Retail 0.602 against 0.599 over eight folds. With per-product class boundaries, the Online Retail models fell to 34–38% accuracy, close to chance. On the cross-sectional BigMart data, logistic regression reached 0.688 accuracy against 0.488 for an outlet-only rule, driven mainly by item price. High classification accuracy therefore mostly reflects product scale and recent sales, not forecasting skill. We also specify a benchmark that compares the serverless endpoint with an always-on endpoint. No latency, cold-start, throughput or cost measurements exist yet, so the paper makes no claim about operational benefits.
KEYWORDS
We proposed SalesSense, an event-driven serverless architecture for product-sales classification and decision support, and specified a benchmark against an always-on baseline. Experiments on three public retail datasets support only preliminary statements about prediction quality. On the temporal datasets, naive baselines matched or beat the learned models, and on Online Retail per-product class boundaries brought every learned model close to chance. On BigMart the models learn real signal, mostly from item price. The evidence does not support claims about latency, elasticity, cold starts, reliability, scalability or cost. The next steps are to deploy the architecture and run the Section VII benchmark, consider target definitions that remove product scale, test on larger cohorts and the full datasets, and release the data provenance and code. No formal hypothesis is tested in this study.
[1] S. Batrel, A. M. Brahme, and J. Gaikwad, "A comprehensive overview of data mining algorithms," Library Progress Int., 2024. [Online]. Available: https://www.researchgate.net/publication/384562785 [2] P. Wadkar and S. D. Mundhe, "Comparative study of different machine learning algorithms used for credit card fraud detection," Int. J. Res. Finance Manage., 2025. [Online]. Available: https://www.researchgate.net/publication/396213597 [3] S. R. Misal, T. K. Kathane, and S. D. Mundhe, "A study on credit card fraud detection using machine learning," 2020. [Online]. Available: https://www.researchgate.net/publication/376833594 [4] S. R. Misal, P. Wadkar, and S. D. Mundhe, "Comparative analysis of different machine learning algorithms used in breast cancer prediction," 2023. [Online]. Available: https://www.researchgate.net/publication/376833481 [5] N. Mandhane and P. Wadkar, "Sentiment analysis of Amazon reviews datasets using machine learning and deep learning algorithms," Int. J. Res. Publ. Rev., 2025. [Online]. Available: https://www.researchgate.net/publication/396697030 [6] N. Pachpor, A. M. Brahme, A. Labade, et al., "Rainfall forecasting and analysis using machine learning models," conference paper, 2025. [Online]. Available: https://www.researchgate.net/publication/403151930 [7] N. P. Ganeshar, "Multiple disease prediction using machine learning," conference paper, 2024. [Online]. Available: https://www.researchgate.net/publication/388743082 [8] T. M. Sonawane and V. N. Ugare, "Docker containers: Revolutionizing software deployment and infrastructure management," conference paper, 2025. [Online]. Available: https://www.researchgate.net/publication/398858555 [9] V. N. Ugare, "Challenges in seamless remote connectivity between on-premise systems and cloud platforms," conference paper, 2026. [Online].Available: https://www.researchgate.net/publication/408539621 [10] S. Batrel, A. M. Brahme, and J. Gaikwad, "An examination of impact of cloud computing and Kubernetes on data availability: An empirical approach," Afr. J. Biol. Sci., 2024. [Online]. Available: https://www.researchgate.net/publication/380574477
Publication Date
10/22/2026
Peer Review Type
Double-Blind Peer Review
Review Rounds: 2
Documents
Article Views
4
Total views since publication
PDF Downloads
7
Total downloads since publication
Geographic Reach
5
Digital Preservation: This article is permanently archived on Zenodo and complies with all data preservation standards across major indexing platforms.
Plagiarism Screening
This manuscript has been checked for plagiarism using TurnitinPlagiarismChecker software.
Conflict of Interest
The authors declare no conflict of interest related to this research work.
Funding Information
No funding information available for this research.
Related research in the same subject area
Loading similar articles...