Social Network Growth & Retention Simulation

Overview

This project involved the design and implementation of a probabilistic agent-based simulation modelling how a social network grows and retains users over time.

Rather than building a functional social platform, the objective was to explore how personality-driven behaviour, stochastic interaction patterns, and retention mechanisms influence network expansion and long-term sustainability.

The system was implemented in Java and complemented with quantitative analysis in Python.

---

Core Objectives

The simulation aimed to investigate:

• how personality traits influence user engagement
• how probabilistic invitation behaviour drives network growth
• how churn mechanisms affect long-term stability
• how engagement scores correlate with retention
• whether different parameter configurations lead to linear, quadratic, or exponential growth

The focus was on modelling system dynamics under uncertainty.

---

Agent-Based Behaviour Model

Each user was represented as an object with:

• randomly generated personality attributes
• probabilistic interaction behaviour
• engagement scoring mechanisms
• invitation likelihood influenced by behavioural state
• churn probability based on engagement and network conditions

Growth emerged from local probabilistic decisions, not deterministic scripts.

This allowed the simulation to produce realistic, non-linear expansion patterns driven by behavioural variability.

---

Retention & Engagement Dynamics

The system explicitly modelled:

• engagement score accumulation
• reduced activity over time
• churn thresholds
• behavioural differences between high- and low-engagement users

Multiple strategies were implemented to compare how different retention assumptions influenced overall growth.

This enabled experimentation with both expansion-focused and stability-focused network dynamics.

---

Graph Representation & Network Visualisation

The social structure was represented as a graph:

• nodes representing individual users
• edges representing friendships

A Java graph visualisation library was used to observe structural evolution over time, including:

• clustering patterns
• network density
• connectivity growth
• fragmentation under churn conditions

No full GUI was implemented, as the emphasis remained on simulation accuracy and analysis.

---

Quantitative Analysis (Python)

Simulation outputs were exported for post-processing in Python.

Analysis included:

• growth curve modelling
• comparison between behavioural strategies
• evaluation of churn impact
• statistical interpretation of engagement dynamics

The results formed the basis of a formal academic dissertation.

---

Timeframe & Context

• Duration: ~9 months
• Context: Master’s degree academic project
• Focus: Probabilistic modelling, network theory, behavioural simulation
• Constraint: Academic reproducibility and analytical rigour

---

Skills Demonstrated

This project demonstrates:

• agent-based modelling under uncertainty
• stochastic system design
• behavioural simulation engineering
• application of design patterns beyond UI contexts
• graph-based system representation
• bridging software implementation and quantitative analysis
• structured academic reporting

---

Why This Project Matters

This project reflects an ability to:

• model complex systems using simple probabilistic rules
• reason about growth and retention dynamics mathematically
• translate abstract behavioural concepts into executable simulations
• analyse emergent behaviour through structured experimentation

It showcases systems thinking applicable to platform engineering, network dynamics, and large-scale user systems.

---