Journal of Intelligent Decision Technologies and Applications (e-ISSN: 3049-0219, p-ISSN: 3139-6569)

AI-based Career Recommendation System for Youth Employment Using Machine Learning

2026-09-18T06:33:17+00:00

Choosing a career is a difficult decision for students because a useful choice depends on more than knowing which occupations are popular. Students may have a reasonable idea of what they enjoy, yet still struggle to translate their abilities into a realistic career direction. This study develops a machine-learning-based career recommendation system that uses structured student skill information to predict a suitable career category. The dataset contains 3,600 student records with attributes representing linguistic ability, logical reasoning, spatial intelligence, and interpersonal skill. The records were prepared through data cleaning, treatment of missing and duplicate entries, categorical encoding, and feature selection before model training. Two supervised classification approaches, Decision Tree and Random Forest, were implemented in Python using Scikit-learn in Google Colab. The data were divided into training and testing subsets using an 80:20 split, and the models were assessed using accuracy, precision, recall, F1-score, and a confusion matrix. The Decision Tree achieved 97.50% accuracy, while Random Forest reached 97.77%. Precision, recall, and F1-score for the Random Forest model were approximately 0.98. The results indicate that the proposed approach can learn a strong relationship between the selected skill attributes and the career labels contained in the dataset. The system is intended as a decision-support tool rather than a replacement for counselors, since career choice also depends on interests, academic history, context, opportunities, and changing labour-market conditions. The work demonstrates a simple and scalable foundation that can be extended with richer student and labour-market information.

Fairlytics: AI-based Intelligent Dynamic Pricing with Discount Authenticity Validation

2026-08-31T11:15:53+00:00

E-commerce platforms increasingly rely on algorithmic pricing to remain competitive, yet two problems persist side by side: prices that fail to track real-time demand, and discount offers that mislead buyers through inflated “strike-through” reference prices, fake festival markdowns, and recycled coupon claims. This article presents Fairlytics, an AI-based intelligent system that unifies dynamic pricing with automated discount authenticity validation in a single pipeline. Fairlytics estimates a fair, demand-responsive price for a product using historical sales, competitor prices, inventory levels, and seasonal signals, and simultaneously verifies whether an advertised discount is genuine by reconstructing the product's true historical price trajectory and comparing it against the claimed original price. A hybrid ensemble of regression and gradient-boosted tree models drives the pricing engine, while a time-series anomaly detector and rule-based authenticity scorer flag manipulated discounts. The proposed architecture is modular, combining data ingestion, feature engineering, dual prediction engines, and a fairness-scoring dashboard for administrators and consumers. This work is intended to improve pricing transparency, protect consumers from deceptive discounting, and give platform operators a defensible, explainable basis for dynamic pricing decisions.

SafeZone: An AI-powered Safety-aware Locality Recommendation Platform for Indian Urban Environments

2026-09-10T12:15:33+00:00

Urban travel and movement have become an important part of everyday life, but selecting a suitable place cannot always depend only on distance, price, convenience, or user ratings. Safety is also an important factor for people travelling alone, families, commuters, and visitors who may not be familiar with a city. However, existing navigation and location-based applications generally emphasize convenience and efficiency, while safety-related information is often treated as a secondary attribute. To address this issue, this paper presents SafeZone, an AI-assisted safety-aware locality recommendation platform designed for Indian urban environments. The proposed system combines locality information, user preferences, safety ratings, and distance to generate ranked recommendations. SafeZone is implemented as a client-side web application and supports six Indian cities—Rajahmundry, Kakinada, Hyderabad, Bangalore, Delhi, and Mumbai—and six categories: hospitals, restaurants, banks, police stations, transport hubs, and tourist spots. A four-step preference-elicitation wizard captures city, category, minimum safety level, minimum rating, maximum budget, ambience, and visit purpose. These preferences are processed through a multi-criteria filter-and-rank mechanism that first applies user-defined constraints and then computes a composite score using rating, safety score, and distance. The SafeZone hybrid strategy was evaluated against keyword search, rule-based filtering, collaborative filtering, and content-based filtering using 1,000 simulated user sessions and 200 expert-annotated query-result pairs. At K = 10, SafeZone achieved a precision of 0.88, recall of 0.85, and F1-score of 0.86, exceeding the evaluated baselines. The findings indicate that combining safety constraints with weighted ranking can produce focused recommendations with low response time at the current database scale.