Aware Personal Finance Management: Memory-Augmented Expense Categorisation with Abstention, Mid-Month Budget-Breach Warning and Per-User Anomaly Scoring
Abstract & Details
Research Area
Computer Science / Artificial Intelligence / Machine Learning
Keywords
Personal finance management
transaction categorisation
selective prediction
conformal prediction
budget forecasting
anomaly detection
budget prediction
human-in-the-loop learning
Abstract
Personal finance applications record spending but rarely adapt to the person using them. Two people can label the same merchant differently, new merchants appear constantly, and a monthly budget is often discovered to be broken only after the month has ended. This paper studies a correction-aware design for an AI budget tracker with four parts: (i) a global text classifier for transaction descriptions that is extended by a per-user memory of corrections; (ii) a confidence threshold that lets the system abstain and ask the user instead of guessing; (iii) a mid-month budget-breach warning that uses spending to date together with history; and (iv) per-user robust anomaly scoring with a cold-start fallback. We evaluate every part on a reproducible synthetic population of 36 users and about 44,055 transactions per run, over five random seeds. A global model reached 70.3±1.7% accuracy on unseen users; adding 50 corrections from a user raised it to 90.0±1.5%. With a confidence threshold of 0.7 the system labelled 89.9% of transactions automatically at 95.5% accuracy. A day-15 early-warning model detected budget breaches with F1 = 0.829 against 0.737 for linear extrapolation, and per-user robust scoring reached an AUC of 0.973 against 0.860 for a global z-score. Conformal forecast intervals under-covered their nominal level under seasonal shift, which we report as a limitation. All tables are regenerated by a single script and are reported as mean ± standard deviation across seeds. All data are synthetic; the results show relative behaviour of the methods under stated assumptions and are not claims about real banking data.
License
This work is licensed under a Creative
Commons
Attribution-ShareAlike 4.0 International License.
Author Information
| # | Name | Institute / Affiliation |
|---|---|---|
| 1 | Pratiksha Yelekar | Guru Nanak Institute of Technology, Nagpur, India |
| 2 | Shruti Selokar | Guru Nanak Institute of Technology, Nagpur, India |
| 3 | Samruddhi katkamwar | Computer Science Engineering Guru Nanak Institute of Technology, Nagpur, India |
| 4 | Jagruti Ghatole | Computer Science Engineering Guru Nanak Institute of Technology, Nagpur, India |
How to Cite
Use the following formats to cite this article in your research.
APA Style
Yelekar, Pratiksha, Selokar, Shruti, katkamwar, Samruddhi, & Ghatole, Jagruti (2026). Aware Personal Finance Management: Memory-Augmented Expense Categorisation with Abstention, Mid-Month Budget-Breach Warning and Per-User Anomaly Scoring. International Journal of Advance Research and Innovative Ideas In Education, 12(5), 639-645.
MLA Style
Yelekar, Pratiksha, et al. "Aware Personal Finance Management: Memory-Augmented Expense Categorisation with Abstention, Mid-Month Budget-Breach Warning and Per-User Anomaly Scoring." International Journal of Advance Research and Innovative Ideas In Education, vol. 12, no. 5, 2026, pp. 639-645.
IEEE Style
Pratiksha Yelekar, Shruti Selokar, Samruddhi katkamwar, and Jagruti Ghatole, "Aware Personal Finance Management: Memory-Augmented Expense Categorisation with Abstention, Mid-Month Budget-Breach Warning and Per-User Anomaly Scoring," International Journal of Advance Research and Innovative Ideas In Education, vol. 12, no. 5, pp. 639-645, 2026.
Vancouver Style
Yelekar Pratiksha, Selokar Shruti, katkamwar Samruddhi, Ghatole Jagruti. Aware Personal Finance Management: Memory-Augmented Expense Categorisation with Abstention, Mid-Month Budget-Breach Warning and Per-User Anomaly Scoring. International Journal of Advance Research and Innovative Ideas In Education. 2026;12(5):639-645.
Harvard Style
Yelekar, Pratiksha, Selokar, Shruti, katkamwar, Samruddhi, & Ghatole, Jagruti (2026) 'Aware Personal Finance Management: Memory-Augmented Expense Categorisation with Abstention, Mid-Month Budget-Breach Warning and Per-User Anomaly Scoring', International Journal of Advance Research and Innovative Ideas In Education, 12(5), pp. 639-645.
Chicago Style
Yelekar, Pratiksha, et al. "Aware Personal Finance Management: Memory-Augmented Expense Categorisation with Abstention, Mid-Month Budget-Breach Warning and Per-User Anomaly Scoring." International Journal of Advance Research and Innovative Ideas In Education 12, no. 5 (2026): 639-645.
Turabian Style
Yelekar, Pratiksha, et al. "Aware Personal Finance Management: Memory-Augmented Expense Categorisation with Abstention, Mid-Month Budget-Breach Warning and Per-User Anomaly Scoring." International Journal of Advance Research and Innovative Ideas In Education 12, no. 5 (2026): 639-645.
Related Research
EduQuest AI: A Gamified Learning Platform for Interactive and Personalized Digital Education
Download PDF
Distributed Real-Time Stream Analytics and Ingestion Engine for High-Velocity Big Data Architectures
Download PDF