snowbasin-traffic-lstm

Model Description

This model predicts traffic counts for Snowbasin ski resort based on time, weather, and historical traffic data.

Model Type

LSTM Neural Network

Performance Metrics

• RMSE: 104.68406516040494
• MAE: 68.94930388211087
• R2: 0.9055348922165798

Features

The model uses the following features:

• Temporal features (hour, day of week, month with cyclical encoding)
• Weather data (temperature, humidity, precipitation, wind speed, snow depth)
• Traffic lag features (1, 2, 3, 6, 12, 24, 168 hours)
• Special features (peak hours, federal holidays, holiday weekends)

Usage

import torch
from huggingface_hub import hf_hub_download

# Download model
model_path = hf_hub_download(repo_id="YOUR_USERNAME/snowbasin-traffic-lstm", filename="champion_lstm.pth")

# Load model architecture (you'll need the TrafficLSTM class)
model = TrafficLSTM(input_size=34)  # Adjust input_size based on your features
model.load_state_dict(torch.load(model_path))
model.eval()

Training Details

• Split Strategy: sequence
• Train Rows: 44600
• Test Rows: 10521

Citation

Developed by Team Wildcats for CS6580 Capstone Project - Spring 2026