{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"While KMeans is a good algorithm, the time complexity is very poor. Kmeans works in O(n⋅K⋅I⋅f) Where n is number of records, K is number of clusters, I is number of iterations, f is number of features in particular record. Clearly, the algorithm will take forever to complete on a dataset of > 100,000 data points"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Minibatch KMeans\n",
"Main features of Minibatch KMeans are:\n",
"\n",
"* Instead of using the entire dataset at once, it operates in batches.\n",
"* Uses Gradient Descent update, which is way more faster than what KMeans does.\n",
"\n",
"#### How it works \n",
"* It takes batches of datasets and finds the centroids for the smaller dataset (minibatch) \n",
"* Then for the next batch, it uses the centroid found in previous batch and updates it using Gradient Descent. \n",
"* This simple method makes it faster by a magnitude of the input size."
]
},
{
"cell_type": "code",
"execution_count": 9,
"metadata": {},
"outputs": [],
"source": [
"import numpy as np\n",
"import matplotlib.pyplot as plt\n",
"import pandas as pd\n",
"import time"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"\n",
"\n",
"
\n",
" \n",
" \n",
" CustomerID \n",
" Genre \n",
" Age \n",
" Annual Income (k$) \n",
" Spending Score (1-100) \n",
" \n",
" \n",
" \n",
" \n",
" 0 \n",
" 1 \n",
" Male \n",
" 19 \n",
" 15 \n",
" 39 \n",
" \n",
" \n",
" 1 \n",
" 2 \n",
" Male \n",
" 21 \n",
" 15 \n",
" 81 \n",
" \n",
" \n",
" 2 \n",
" 3 \n",
" Female \n",
" 20 \n",
" 16 \n",
" 6 \n",
" \n",
" \n",
" 3 \n",
" 4 \n",
" Female \n",
" 23 \n",
" 16 \n",
" 77 \n",
" \n",
" \n",
" 4 \n",
" 5 \n",
" Female \n",
" 31 \n",
" 17 \n",
" 40 \n",
" \n",
" \n",
"
\n",
"
"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"fig = plt.figure(figsize=(15,6))\n",
"plt.subplot(121)\n",
"plt.scatter(X[:, 0], X[:, 1], c=y_minikmeans, s=50, cmap='viridis')\n",
"\n",
"centers = clf.cluster_centers_\n",
"plt.scatter(centers[:, 0], centers[:, 1], c='black', s=150, alpha=0.5)\n",
"plt.title('Clusters using DBSCAN')\n",
"plt.ylabel('Grocery')\n",
"plt.xlabel('Milk')\n",
"\n",
"plt.subplot(122)\n",
"plt.scatter(X[:, 0], X[:, 1], c=y_kmeans, s=50, cmap='viridis_r')\n",
"\n",
"centers = kmean.cluster_centers_\n",
"plt.scatter(centers[:, 0], centers[:, 1], c='black', s=150, alpha=0.5)\n",
"plt.title('Clusters using kmeans')\n",
"plt.ylabel('Grocery')\n",
"plt.xlabel('Milk')\n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"well there is not any difference between both for small datasets"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Disad\n",
"The mini batch K-means is faster but gives slightly different results than the normal batch K-means."
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.7"
}
},
"nbformat": 4,
"nbformat_minor": 2
}