implemented spectral clustering

tjresearch · Jun 17, 2020 · 3fdeabd · 3fdeabd
1 parent 960e663
commit 3fdeabd
Showing 1 changed file with 32 additions and 1 deletion.
diff --git a/multitask/Cluster.ipynb b/multitask/Cluster.ipynb
@@ -16,7 +16,10 @@
  "outputs": [],
  "source": [
  "import itertools\n",
- "import pandas as pd"
+ "import pandas as pd\n",
+ "import pickle\n",
+ "\n",
+ "from sklearn.cluster import SpectralClustering"
  ]
  },
  {
@@ -124,6 +127,24 @@
  " return df"
  ]
  },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "def cluster_from_similarity_matrix(df, num_clusters):\n",
+ " mat = df.to_numpy()\n",
+ " df[:] = (mat+mat.T)/2\n",
+ " sc = SpectralClustering(n_clusters=num_clusters, affinity='precomputed')\n",
+ " labels = sc.fit_predict(df.to_numpy())\n",
+ " d = {c:[] for c in labels}\n",
+ " for c, t in zip(labels, df.columns):\n",
+ " d[c].append(t)\n",
+ " clusters = [v for k, v in d.items()]\n",
+ " return clusters"
+ ]
+ },
  {
  "cell_type": "code",
  "execution_count": null,
@@ -141,6 +162,16 @@
  "source": [
  "save_transfer_coefficients(tasks, A_test, X_test, E_test, y_test)"
  ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "df = load_transfer_coefficients()\n",
+ "print(cluster_from_similarity_matrix(df, 6))"
+ ]
  }
  ],
  "metadata": {