GeoIMC Tutorial: Geographic Inductive Matrix Completion

GeoIMC Tutorial: Geographic Inductive Matrix Completion#

Sandbox Model — Experimental

This model is part of the CoreRec Sandbox — an experimental collection of algorithms for research and exploration. Sandbox models are not production-tested and may have incomplete implementations. For production-ready models, see the Production Models.

Introduction#

GeoIMC leverages geographic information and graph structure for location-based recommendations using inductive matrix completion.

Paper: Spatial Matrix Completion

How GeoIMC Works#

Architecture#

Spatial-Aware Matrix Completion:

  1. Geographic Graph: Locations connected by proximity

  2. Graph Convolution: Propagate location features

  3. User Preferences: Latent factors

  4. Inductive Completion: Generalize to new locations

  5. Spatial Regularization: Nearby locations similar

Key Innovation: Combines CF with geographic proximity

Architecture:

Location Graph → GCN → Location Embeddings                                             → MF → Rating Prediction
User Features  → MLP → User Embeddings     /

Mathematical Foundation#

Graph Convolution on Locations:

h_l^(k+1) = σ(Σ_{l'∈N(l)} W^k · h_l'^(k) / |N(l)|)

Matrix Factorization:

r̂_ul = p_u^T · q_l + b_u + b_l

Spatial Regularization:

L_spatial = λ_s · Σ_{l,l'∈neighbors} ||q_l - q_l'||²

Total Loss:

L = L_MF + L_spatial + L_graph

Reference Implementation#

Implementation Note

The code examples below are reference implementations for learning purposes. This is a sandbox model and has not been production-tested. Verify behavior thoroughly before using in production.

Step 1: Import and Load Data#

from corerec.sandbox.collaborative_full.graph_based_base.geoimc import GeoIMC
from cr_learn import ml_1m
from sklearn.model_selection import train_test_split
import numpy as np

# Load dataset with CORRECT API
data = ml_1m.load()  # Returns dict with 'ratings', 'users', 'movies'
ratings_df = data['ratings']  # DataFrame with user_id, movie_id, rating, timestamp

print(f"Loaded {len(ratings_df)} ratings")

# Split data
train_df, test_df = train_test_split(ratings_df, test_size=0.2, random_state=42)

# Extract arrays for model
train_users = train_df['user_id'].values
train_items = train_df['movie_id'].values
train_ratings = train_df['rating'].values

test_users = test_df['user_id'].values
test_items = test_df['movie_id'].values
test_ratings = test_df['rating'].values

Step 2: Initialize Model#

model = GeoIMC(
    name="GeoIMC_Model",
    n_factors=64,
    gcn_layers=2,
    spatial_reg=0.5,
    neighbor_radius_km=2.0,
    epochs=20,
    batch_size=256,
    learning_rate=0.001,
    verbose=True
)

print(f"Initialized {model.name}")

Step 3: Train#

model.fit(
    user_ids=train_users,
    item_ids=train_items,
    ratings=train_ratings
)

print("Training complete!")

Step 4: Predict#

# Single prediction
score = model.predict(user_id=1, item_id=100)
print(f"Predicted score: {score:.3f}")

# Batch predictions
test_predictions = model.batch_predict(list(zip(test_users[:100], test_items[:100])))

Step 5: Recommend#

# Get top-10 recommendations for user
user_id = 1
recommendations = model.recommend(
    user_id=user_id,
    top_k=10
)

print(f"Top-10 recommendations for User {user_id}:")
for rank, item_id in enumerate(recommendations, 1):
    print(f"  {rank}. Item {item_id}")

Step 6: Evaluate#

from sklearn.metrics import mean_squared_error
import numpy as np

# Predict all test ratings
test_pred = [model.predict(u, i) for u, i in zip(test_users, test_items)]
rmse = np.sqrt(mean_squared_error(test_ratings, test_pred))
print(f"Test RMSE: {rmse:.4f}")

Step 7: Save & Load#

# Save model
model.save('artifacts/geoimc_model')

# Load model
loaded = GeoIMC.load('artifacts/geoimc_model')
test_score = loaded.predict(1, 100)
print(f"Loaded model prediction: {test_score:.3f}")

Key Takeaways#

When to Use GeoIMC#

✅ Perfect For:

  • Point-of-Interest (POI) recommendation

  • Location-based services (Yelp, Foursquare)

  • Check-in data

  • When geography matters

  • Cold-start locations (inductive)

❌ Not For:

  • No geographic structure

  • Pure collaborative filtering

  • When location doesn’t matter (movies, books)

  • Very sparse check-ins (<100 per user)

Best Practices#

  1. Graph Construction: k-NN or radius-based (1-5km)

  2. GCN Layers: 2-3 layers

  3. MF Factors: 50-100

  4. Spatial Regularization: λ_s = 0.1-1.0

  5. Distance Weighting: Inverse distance or Gaussian

  6. Features: Include category, popularity

  7. Negative Sampling: Geographic-aware sampling

  8. Train/Test: Split by time, not random

Scaling & Production Considerations#

If you plan to move this sandbox model toward production use, consider:

  • Data Volume: Test with realistic dataset sizes; sandbox implementations may not be optimized for large-scale data.

  • Distributed Training: For datasets exceeding single-machine memory, consider wrapping the model in distributed training frameworks (e.g., PyTorch DDP, Horovod).

  • Hyperparameter Tuning: Use systematic search (Optuna, Ray Tune) rather than manual tuning. The defaults in sandbox models are not tuned for any specific domain.

  • Serving Latency: Profile inference time. Some architectures (e.g., attention-based) may need quantization or ONNX export for low-latency serving.

  • Monitoring: Implement A/B testing and online metrics (CTR, engagement, diversity) before replacing an existing system.

  • Validation: Write comprehensive unit tests and integration tests before deploying. Sandbox models do not have CI coverage.

Further Reading#

Contents