Mini Batch K-Means

Fast variant of K-Means that uses mini-batches of data to reduce computation time.

When to use:

• Very large datasets (>10k samples)
• Need faster training than K-Means
• Can accept slightly lower quality for speed
• Memory constraints

Strengths: Much faster than K-Means, lower memory usage, good for large datasets, similar quality to K-Means Weaknesses: Slightly less accurate than K-Means, more sensitive to initialization

Model Parameters

N Clusters (default: 8, required) Number of clusters to form.

Init Method (default: "k-means++") How to initialize cluster centers:

• k-means++: Smart initialization (better)
• random: Random initialization (faster)

Max Iterations (default: 100) Maximum iterations over the complete dataset.

• 50-100: Usually sufficient for large data
• 200+: For better convergence

Batch Size (default: 1024) Size of mini batches for training.

• 256-512: Small batches (more updates, slower)
• 1024: Good default
• 2048+: Large batches (fewer updates, faster)

Random State (default: 42) Seed for reproducibility.