HappyFest: An AI-Assisted Web Platform for Event Planning, Vendor Recommendation, and Budget Allocation Using Random Forest and Sentiment Analysis

This paper presented HappyFest, a role-based Flask web application that unifies event creation, ML-driven vendor-category recommendation, rule-based budget allocation, vendor booking and feedback-driven sentiment analysis on a single platform for home and small-institutional event planning — a user group whose planning workflow existing event-recommendation research largely leaves on the attendee-facing side of the problem. A RandomForestClassifier trained on a compact, curated 37-record event–vendor dataset supplies a fast, explainable vendor-category recommendation with a confidence score, while a transparent, event-type-aware rule-based engine turns the same event parameters into a seven-category budget breakdown, a per-guest cost figure and AI-generated cost-saving insights. A TextBlob/lexicon-based sentiment-analysis routine converts free-text post-event feedback into a structured Positive/Neutral/Negative signal without any separate batch process, feeding directly into an administrator analytics dashboard. The application was implemented across all three intended roles — User, Organizer and Administrator — and tested end-to-end; a diagnostic test script together with fourteen representative, manually exercised system test cases confirm that registration, event management, recommendation and budget prediction, vendor booking, feedback sentiment analysis and administrative analytics all function correctly together. The project also demonstrates that meaningful decision support does not always require the heaviest available model: a modest, fast-training Random Forest classifier and a transparent rule-based budget engine were sufficient to produce useful, explainable recommendations given the limited historical data available to a new platform, while keeping the system inexpensive to run and easy to reason about for both developers and evaluators. Limitations: Two limitations are worth stating plainly. First, the recommendation model is trained on a deliberately small, curated 37-record dataset rather than on real historical booking data at scale, so its recommendations reflect the priors encoded in that seed dataset rather than learned, platform-specific user behaviour; Section 2’s review of attendance-prediction and recommendation literature [1]–[3] suggests that accuracy and personalisation typically improve substantially once real usage data becomes available. Second, the sentiment-analysis module relies on TextBlob’s lexicon-and-pattern-based scoring, which — consistent with the limitations reported by Hazarika et al. [7] and Prasanna Kumar [8] — struggles with sarcasm, negation and domain-specific phrasing in short feedback comments. Both limitations are addressed directly in the future work below.

[1] D. Horowitz, D. Contreras, and M. Salamó, “EventAware: A Mobile Recommender System for Events,” Pattern Recognition Letters, vol. 105, pp. 121–134, 2017. [2] R. Varaprasad, S. Ramasubbareddy, and K. Govinda, “Event Recommendation System Using Machine Learning Techniques,” in Innovations in Computer Science and Engineering, Lecture Notes in Networks and Systems, vol. 385, Springer, Singapore, 2022. [3] R. Awad, “Event Attendance Prediction and Recommendation System: A Comparative Study,” M.S. thesis, Politecnico di Milano, 2023. [4] “EventVerse: A Smart Web-Based Event Planning and Vendor Recommendation System Using Supabase and React,” International Journal of Computer Techniques (IJCT), vol. 13, no. 3, 2026. [5] L. Breiman, “Random Forests,” Machine Learning, vol. 45, no. 1, pp. 5–32, 2001. [6] F. Ricci, L. Rokach, and B. Shapira, Eds., Recommender Systems Handbook, 2nd ed., Springer, Boston, MA, 2015. [7] D. Hazarika, G. Konwar, S. Deb, and D. J. Bora, “Sentiment Analysis on Twitter by Using TextBlob for Natural Language Processing,” Proc. Int. Conf. on Research in Management & Technovation, ACSIS, vol. 24, pp. 63–67, 2020. [8] Prasanna Kumar H. R., “Sentiment Analysis: TextBlob for Decision Making,” International Journal of Scientific Research & Engineering Trends, 2023. [9] Pallets Projects, “Flask Documentation,” https://flask.palletsprojects.com/. [10] scikit-learn developers, “scikit-learn: Machine Learning in Python,” https://scikit-learn.org/.