[DOC] Notebook on Feature-based Clustering (aeon-toolkit#2579)

* Feature-based clustering

* Feature-based clustering update

* Update clustering overview

* formatting

Author: itsdivya1309

examples/clustering/clustering.ipynb

@@ -2,6 +2,12 @@
  "cells": [
   {
    "cell_type": "markdown",
+   "metadata": {
+    "collapsed": false,
+    "jupyter": {
+     "outputs_hidden": false
+    }
+   },
    "source": [
     "# Time Series Clustering\n",
     "\n",
@@ -23,45 +29,47 @@
     "erative [16], Feature K-means [17], Feature K-medoids [17], U-shapelets [18],\n",
     "USSL [19], RSFS [20], NDFS [21], Deep learning and dimensionality reduction\n",
     "approaches see [22]"
-   ],
-   "metadata": {
-    "collapsed": false
-   }
+   ]
   },
   {
    "cell_type": "markdown",
+   "metadata": {
+    "collapsed": false,
+    "jupyter": {
+     "outputs_hidden": false
+    }
+   },
    "source": [
     "## Clustering notebooks\n",
     "\n",
-    "- `aeon` currently focusses on partition based approaches that use elastic distance\n",
-    "functions. The [partition based](partitional_clustering.ipynb) note book has an\n",
-    "overview of the funtionality in aeon.\n",
+    "`aeon` offers a comprehensive suite of time series clustering (TSCL) algorithms, encompassing partition-based, density-based, hierarchical, deep learning, and feature-based approaches.\n",
+    "\n",
+    "- `aeon` has many partition-based clustering algorithms, which include TimeSeriesKMeans, KMedoids, CLARA, CLARANS, ElasticSOM, and KSpectralCentroid, leveraging elastic distance measures like DTW. The [partition-based](partitional_clustering.ipynb) notebook has an overview of the functionality in aeon.\n",
     "\n",
-    "- `sklearn` has *density based* and *hierarchical based* clustering algorithms, and\n",
-    "these can be used in conjunction with `aeon` elastic distances. See the [sklearn and\n",
-    "aeon distances](../distances/sklearn_distances.ipynb) notebook.\n",
+    "- `sklearn` has *density-based* and *hierarchical based* clustering algorithms, which can be used in conjunction with `aeon` elastic distances. See the [sklearn and aeon distances](../distances/sklearn_distances.ipynb) notebook.\n",
     "\n",
-    "- Deep learning based TSCL is a very popular topic, and we are working on bringing\n",
-    "deep learning functionality to `aeon`, first algorithms for [Deep learning] are\n",
-    "COMING SOON\n",
+    "- Bespoke feature-based TSCL algorithms are easily constructed with `aeon` transformers and `sklearn` clusterers in a pipeline. Some examples are in the\n",
+    "[sklearn clustering]. The [feature-based](feature_based_clustering.ipynb) notebook gives an overview of the feature-based clusterers in an aeon.\n",
     "\n",
-    "- Bespoke feature based TSCL algorithms are easily constructed with `aeon`\n",
-    "transformers and `sklearn` clusterers in a pipeline. Some examples are in the\n",
-    "[sklearn clustering]. We will bring the bespoke feature\n",
-    "based clustering algorithms into `aeon` in the medium term.\n",
+    "- Deep learning based TSCL is a very popular topic, and we have introduced many deep learning functionalities to `aeon`. Autoencoder-based models like AEFCNClusterer, AEResNetClusterer, and AEDCNNClusterer enable complex pattern discovery.\n",
     "\n",
-    "We are in the process of extending the bake off described in [1] to include all\n",
-    "clusterers. So far, we find medoids with MSM distance is the best performer.\n",
+    "- `aeon` also includes averaging-based clustering algorithms, which utilize centroid-based representations of time series.\n",
+    "\n",
+    "\n",
+    "We are in the process of extending the bake-off described in [1] to include all clusterers. So far, we find medoids with MSM distance is the best performer.\n",
     "\n",
     "<img src=\"img/clst_cd.png\" width=\"600\" alt=\"cd_diag\">\n",
     "\n"
-   ],
-   "metadata": {
-    "collapsed": false
-   }
+   ]
   },
   {
    "cell_type": "markdown",
+   "metadata": {
+    "collapsed": false,
+    "jupyter": {
+     "outputs_hidden": false
+    }
+   },
    "source": [
     "## References\n",
     "\n",
@@ -141,31 +149,28 @@
     "[22] B. Lafabregue, J. Weber, P. Gancarski, and G. Forestier. End-to-end deep\n",
     "representation learning for time series clustering: a comparative study. Data Mining\n",
     "and Knowledge Discovery, 36:29—-81, 2022\n"
-   ],
-   "metadata": {
-    "collapsed": false
-   }
+   ]
   }
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "Python 3",
+   "display_name": "Python 3 (ipykernel)",
    "language": "python",
    "name": "python3"
   },
   "language_info": {
    "codemirror_mode": {
     "name": "ipython",
-    "version": 2
+    "version": 3
    },
    "file_extension": ".py",
    "mimetype": "text/x-python",
    "name": "python",
    "nbconvert_exporter": "python",
-   "pygments_lexer": "ipython2",
-   "version": "2.7.6"
+   "pygments_lexer": "ipython3",
+   "version": "3.11.10"
   }
  },
  "nbformat": 4,
- "nbformat_minor": 0
+ "nbformat_minor": 4
 }