RESEARCH-ARTICLE

Young Tourist Preference Interpretation and Recommendation Optimization for Historic Villages--Taking Dongguan Agarwood Culture as a Core Case Using LightGBM-SHAP

This pioneering research introduces an interpretable machine learning framework, integrating LightGBM with SHAP (SHapley Additive exPlanations) analysis, to accurately predict young tourist preferences and generate actionable optimization strategies for historic villages. Leveraging 20,347 samples from major Chinese travel platforms, the framework identifies critical drivers like accommodation design style and social check-in points, outperforming traditional recommendation systems and achieving 87.6% accuracy and a 250% ROI for key renovation strategies. This data-driven approach bridges predictive analytics with practical rural revitalization, offering transparent insights for youth-oriented destination transformation.

Schedule Your AI Strategy Session

Executive Impact Summary

This study delivers critical advancements for revitalizing historic villages by precisely understanding and catering to young tourist preferences. The integration of advanced AI delivers not only superior predictive accuracy but also unprecedented interpretability, translating directly into highly effective, ROI-positive development strategies.

0% Prediction Accuracy

0 AUC-ROC Score

0% Outperformance vs. CF

0% Max Strategy ROI

Unlock Your Enterprise's Potential

Deep Analysis & Enterprise Applications

Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.

Superior Predictive Performance with LightGBM

Our proposed LightGBM model significantly outperforms traditional recommendation systems, achieving an 87.6% accuracy and a 0.912 AUC score. This robust performance is crucial for precise, personalized recommendations in niche destinations like historic villages. The model's efficiency stems from its histogram-based algorithm and leaf-wise growth strategy, which effectively handle sparse data and high-cardinality features common in tourism datasets.

Model Performance Comparison

Model	Accuracy	AUC
Collaborative Filtering (CF)	78.3%	0.812
Random Forest (RF)	82.1%	0.856
XGBoost	85.2%	0.891
LightGBM (Proposed)	87.6%	0.912

Unveiling Core Drivers with SHAP Explanations

Using SHAP (SHapley Additive exPlanations), we identified the most influential features driving young tourists' preferences, moving beyond "black-box" predictions to actionable insights. The global interpretation ranks these features by their mean absolute SHAP values, providing a clear roadmap for targeted development.

Top 5 Critical Features and Their Impact:

Accommodation Design Style (SHAP value=0.23): Modern minimalist, industrial, and Japanese wabi-sabi styles significantly increase interest, reflecting young consumers' pursuit of "aesthetic economy" and "spatial sociality."
Social Check-in Points (SHAP value=0.19): Villages with 3-5 Instagram-worthy spots are optimal, increasing attraction by 34% and serving as "shareable content." Excessive points lead to "check-in fatigue."
Cultural Product Diversity (SHAP value=0.17): Richness in cultural creative businesses (hand-dyeing, wood carving) positively correlates with interest, shifting preference from "sightseeing" to "experiential tourism."
Night Activity Rating (SHAP value=0.15): Light shows, bonfire concerts, and stargazing camping strongly attract young tourists, extending stay duration and creating differentiated experiences.
Transportation Convenience (SHAP value=0.12): Accessibility within 2 hours from high-speed rail stations is critical, aligning with "weekend micro-vacation" travel patterns.

Key Non-Linear Interactions:

Accommodation Price: Modern minimalist styles achieve peak SHAP values in the 300-500 CNY/night range, representing the "quality consumption sweet spot" and avoiding both "quality concerns" (low price) and "value-for-money skepticism" (high price).
Check-in Point Quantity: An inverted U-shaped relationship shows 3-5 points are optimal, balancing "exploration surprise" with "check-in fatigue" and "superficial experiences" from too many points.

Validated Optimization Strategies for Rural Revitalization

Based on our SHAP insights, we designed and validated three data-driven optimization strategies through counterfactual simulation, demonstrating significant improvements in high-interest sample proportions and strong ROI.

Strategy A: Social Check-in Installation Enhancement

Description: Add 2 Instagram-worthy check-in points (e.g., rainbow stairs, sky mirror), increasing the "check-in point quantity" feature. This strategy saw an 11.2 percentage point increase in high-interest samples.

Strategy B: Cultural Creative Business Introduction

Description: Introduce 5 cultural creative shops (hand-dyeing, wood carving experiences), raising the "cultural product diversity" score. This resulted in a 7.8 percentage point increase in high-interest samples.

Strategy C: Night Economy Development

Description: Develop night light shows and concert projects, increasing the "night activity rating." This led to a 9.3 percentage point increase in high-interest samples.

Cost-Benefit Comparison of Optimization Strategies

Strategy	Initial Investment	Expected Annual Revenue	ROI	Payback Period
Strategy A	100,000 CNY	300,000 CNY	250%	4 months
Strategy B	250,000 CNY	350,000 CNY	108%	11 months
Strategy C	400,000 CNY	500,000 CNY	91%	14 months

Recommended Approach: A combination of Strategy A + Strategy B is recommended for optimal impact, leveraging rapid traffic acquisition from check-in points and deeper experiential value from cultural creative shops, leading to a synergistic effect that boosts high-interest samples by 57.8%.

Dongguan Agarwood Culture: A Localized Success Case

The study specifically validates its strategies using Dongguan agarwood culture as a core local case. This regional cultural symbol is integrated into diverse creative experiences such as agarwood making workshops, cultural heritage tours, and themed activities. These immersive practices enrich tourism products and align with the "cultural experience depth" indicator.

For young tourists, hands-on cultural experiences effectively enhance emotional resonance with the destination, proving that integrating regional characteristic culture into tourism recommendation systems better meets personalized and cultural needs, thereby enhancing the practical value of the research outcomes.

LightGBM Model Accuracy

87.6% Prediction Accuracy for Youth Tourist Preferences

Enterprise Process Flow

Data Collection

→

Feature Engineering (22 dimensions)

→

LightGBM Model Training

→

SHAP Interpretation Analysis

→

Optimization Strategies

Top Feature Drivers for Youth Tourist Preference

Feature	Mean \|SHAP Value\|
Accommodation Design Style	0.23
Social Check-in Points	0.19
Cultural Product Diversity	0.17
Night Activity Rating	0.15
Transportation Convenience	0.12

ROI from Social Check-in Enhancement

250% Return on Investment for Check-in Installation Enhancement

Localized Application: Dongguan Agarwood Culture

The study verifies the practical applicability of optimized strategies by taking Dongguan agarwood culture as a core local case. Integrating regional characteristic culture into tourism recommendations better meets the personalized and cultural needs of young tourist groups, enhancing the practical value of research outcomes, and driving sustainable rural revitalization.

Discuss Regional Strategy

Calculate Your Potential AI-Driven ROI

Estimate the financial and operational benefits of implementing AI solutions tailored to your enterprise needs. Adjust the parameters below to see an instant impact.

Industry Sector

Number of Employees (Impacted by AI)

Avg. Hours per Week on Repetitive Tasks

Average Hourly Wage ($)

Estimated Annual Savings $0

Hours Reclaimed Annually 0

Deep Dive into Your ROI

Your AI Implementation Roadmap

A structured approach ensures successful integration and maximum impact. Our proven methodology guides your enterprise through every phase of AI adoption.

Phase 1: Discovery & Strategy Alignment

Comprehensive analysis of current workflows, identification of AI opportunities, and strategic alignment with business objectives. Define key metrics and success criteria.

Phase 2: Data Preparation & Model Development

Collection, cleaning, and engineering of relevant data. Custom model training and validation, focusing on explainability and performance tailored to your specific use cases.

Phase 3: Pilot Implementation & Optimization

Deployment of AI solutions in a controlled environment. Iterative testing, performance monitoring, and fine-tuning based on real-world feedback to achieve optimal results.

Phase 4: Full-Scale Integration & Training

Seamless integration of validated AI systems into your existing enterprise infrastructure. Comprehensive training for your teams to ensure effective adoption and utilization.

Phase 5: Continuous Monitoring & Evolution

Ongoing performance monitoring, regular updates, and adaptive evolution of AI models to maintain peak efficiency and adapt to changing business needs and market dynamics.

Start Your AI Journey

Ready to Transform Your Enterprise with AI?

Our experts are ready to help you navigate the complexities of AI implementation. Schedule a personalized consultation to explore how these insights can drive tangible results for your business.

Book Your Free Consultation