2024-10-30

Problem Description

Problem to Investigate:

  • Use National Park Service (NPS) API data to optimize visitor experiences.
  • Leverage data on events, facilities, alerts, and news.
  • Goal: Improve satisfaction, resource management, and safety.

Why It’s Interesting:

  • Visitor satisfaction is key for park management.
  • Data identifies visitor patterns, popular events, and alert impacts (e.g., weather).
  • Enables resource allocation, safety improvement, and better experiences.
  • Helps parks adapt to seasonal changes and manage resources effectively.

Analytics Plan

Analyzing the Data:

  1. Data Collection & Preprocessing:
    • Collect data from NPS API on park activities, amenities, services, and location.
    • Clean data, handle missing values, one-hot encode categorical variables, and scale numerical features.
  2. Clustering Analysis (H2O):
    • Use k-means clustering to group parks with similar offerings, revealing patterns in activities and amenities.
    • Identify regional or thematic groupings in visitor experiences.
  3. Classification Modeling (H2O):
    • Build machine learning models (Random Forests, GBMs) to predict visitor satisfaction based on park attributes.
    • Identify features driving high visitor satisfaction to guide NPS resource prioritization.

Methods and Tools:

  • Clustering: H2O for k-means.
  • Classification: H2O models (Random Forests, GBMs) for visitor satisfaction prediction.

Evaluation Plan

Clustering Evaluation (H2O):

✔ Metrics: Silhouette Score.
✔ Purpose: Use H2O’s k-means clustering to group parks based on similar offerings (activities, amenities, services). This reveals patterns in visitor experiences and helps categorize parks by their visitor engagement profiles.

Classification Evaluation (H2O):

✔ Metrics: Accuracy, F1-score, AUC.
✔ Cross-Validation: Apply k-fold validation in H2O for model robustness.
✔ Purpose: Use H2O models (e.g., Random Forests, GBMs) to classify parks as “high-satisfaction” or “low-satisfaction” based on park attributes (e.g., activities, amenities). This predicts satisfaction levels to inform resource prioritization.

Feature Importance Analysis (H2O):

✔ Method: Use SHAP values in H2O to identify key attributes influencing visitor satisfaction, such as specific activities or amenities.
✔ Outcome: Provides insights to guide the National Park Service (NPS) in optimizing resources and enhancing visitor experiences based on data-driven priorities.