ABSTRACT
Online dating applications have revolutionized how individuals initiate and form relationships in today’s digital society. With vast amounts of user behavior data generated on these platforms, machine learning techniques can be leveraged to predict match outcomes and improve user experience. This project focuses on developing a predictive model to forecast dating app match results based on user profile attributes and interaction behaviors. The model aims to analyze key features such as gender, location type, app usage duration, likes received, mutual matches, and the number of messages sent. A structured dataset representing user behaviors was used to train the model. The target variable, match_outcome, was transformed using label encoding to facilitate multi-class classification. Categorical data was encoded, and numerical features were scaled using standardization to enhance model learning. Since the data was imbalanced, RandomOverSampler was applied to balance the target classes and avoid biased predictions. A Random Forest Classifier was trained due to its reliability in handling high-dimensional and non-linear data. To ensure usability, a Tkinter-based graphical interface was developed where users can input features and get instant predictions on match outcomes. The GUI bridges the technical model and user interaction, simulating a real-world deployment scenario. The system demonstrates how AI can be effectively integrated into dating platforms for smarter matchmaking. It also showcases the potential of machine learning in understanding user preferences and optimizing social interactions. This work lays a foundation for future enhancements such as personalized recommendations, adaptive learning, and deep behavioral analytics.

OBJECTIVES
a) Develop a Predictive Model for Match Outcome
To design and implement a machine learning model that predicts the outcome of a match (e.g., mutual match, no match, or conversation initiated) based on user behavior and profile data, including features such as gender, location, app usage time, likes received, mutual matches, and message sent count.
b) Handle Imbalanced Dataset
To address class imbalance in the dataset by implementing techniques such as RandomOverSampling, ensuring that the model learns from all classes equally and provides unbiased predictions for all match outcomes.
c) Implement Feature Engineering for Behavioral Data
To identify and select key features from user behavioral data (e.g., message frequency, swipe behavior, time spent on the app) that most strongly correlate with successful match outcomes and improve model performance.
d) Build a User-Friendly Interface for Predictions
To create a user-friendly graphical interface (GUI) using Tkinter that allows users to input their data (e.g., gender, location, likes, messages) and receive real-time predictions for their match outcome.
e) Evaluate Model Performance with Metrics
To assess the performance of the predictive model using appropriate evaluation metrics such as accuracy, F1 score, precision, and recall, ensuring the model's robustness in predicting match outcomes for a diverse set of users.
f) Save and Deploy Model for Real-Time Use
To save the trained machine learning model and its necessary components (scalers and encoders) using joblib, enabling the deployment of the model for real-time predictions in the GUI application, and ensuring scalability for future enhancements.