RQ2_clusters.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "52416957-ea99-4a94-8f72-0bbf8b174bef",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-02-02T06:27:32.879452Z",
     "iopub.status.busy": "2023-02-02T06:27:32.878814Z",
     "iopub.status.idle": "2023-02-02T06:27:33.185606Z",
     "shell.execute_reply": "2023-02-02T06:27:33.185209Z",
     "shell.execute_reply.started": "2023-02-02T06:27:32.879313Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>name</th>\n",
       "      <th>dependency</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>hammer-sh</td>\n",
       "      <td>tensorflow</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>dlomix</td>\n",
       "      <td>tensorflow</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>artap</td>\n",
       "      <td>tensorflow</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>wikirec</td>\n",
       "      <td>tensorflow</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>hipomap</td>\n",
       "      <td>tensorflow</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        name  dependency\n",
       "0  hammer-sh  tensorflow\n",
       "1     dlomix  tensorflow\n",
       "2      artap  tensorflow\n",
       "3    wikirec  tensorflow\n",
       "4    hipomap  tensorflow"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "from pymongo import MongoClient\n",
    "\n",
    "tf_edges = pd.read_csv(\n",
    "    \"data/TensorFlow/TensorFlow_edges.csv\", names=[\"name\", \"dependency\"]\n",
    ")\n",
    "pt_edges = pd.read_csv(\"data/PyTorch/PyTorch_edges.csv\", names=[\"name\", \"dependency\"])\n",
    "tf_edges.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "6c0361fa-ab79-4412-9000-a82d41b388bf",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-09-02T04:22:31.298801Z",
     "iopub.status.busy": "2022-09-02T04:22:31.298658Z",
     "iopub.status.idle": "2022-09-02T04:22:31.447800Z",
     "shell.execute_reply": "2022-09-02T04:22:31.447108Z",
     "shell.execute_reply.started": "2022-09-02T04:22:31.298784Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import igraph as ig\n",
    "import leidenalg as la\n",
    "import networkx as nx\n",
    "\n",
    "\n",
    "def get_communities(framework: str, package_names: list, seed=None):\n",
    "    vertices = []\n",
    "    edges = []\n",
    "    with open(f\"data/{framework}/{framework}_edges.csv\") as f:\n",
    "        for line in f:\n",
    "            down_pkg, up_pkg = line.strip(\"\\n\").split(\",\")\n",
    "            has_layer1_pkg = (down_pkg in package_names) or (up_pkg in package_names)\n",
    "            if not has_layer1_pkg:\n",
    "                vertices.append(down_pkg)\n",
    "                vertices.append(up_pkg)\n",
    "                edges.append((up_pkg, down_pkg))\n",
    "    vertices = list(set(vertices))\n",
    "    nx_dig = nx.DiGraph()\n",
    "    nx_dig.add_edges_from(edges, weight=1)\n",
    "    edges = [(vertices.index(u), vertices.index(d)) for u, d in edges]\n",
    "    ig_dig = ig.Graph(directed=True)\n",
    "    ig_dig.add_vertices(len(vertices))\n",
    "    ig_dig.add_edges(edges)\n",
    "    partition = la.find_partition(\n",
    "        ig_dig, la.ModularityVertexPartition, n_iterations=2, seed=seed\n",
    "    )\n",
    "    comm_node_centrality = []\n",
    "    for par in partition:\n",
    "        subg = nx_dig.subgraph([vertices[v] for v in par])\n",
    "        centrality = nx.katz_centrality(subg.reverse())\n",
    "        centrality = sorted(centrality.items(), key=lambda x: x[0], reverse=True)\n",
    "        comm_node_centrality.append(\n",
    "            sorted(centrality, key=lambda x: x[1], reverse=True)\n",
    "        )\n",
    "\n",
    "    with open(f\"data/{framework}/{framework}_communities\", \"w\") as outf:\n",
    "        tmp = sorted(comm_node_centrality, key=lambda x: x[0][0].lower())\n",
    "        tmp = sorted(tmp, key=lambda x: len(x), reverse=True)\n",
    "        for comm in tmp:\n",
    "            nodes = [v[0] for v in comm]\n",
    "            outf.write(\", \".join(nodes))\n",
    "            outf.write(\"\\n\")\n",
    "    return nx_dig, partition, comm_node_centrality"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "9fa60be8-0b7b-408b-bce4-48599baf3f12",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-09-02T04:22:31.449346Z",
     "iopub.status.busy": "2022-09-02T04:22:31.449113Z",
     "iopub.status.idle": "2022-09-02T04:22:31.457456Z",
     "shell.execute_reply": "2022-09-02T04:22:31.456747Z",
     "shell.execute_reply.started": "2022-09-02T04:22:31.449326Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def comm_pkg_info(comm: list):\n",
    "    info_list = []\n",
    "    for pkg in comm:\n",
    "        info_list.append([pkg, f\"https://pypi.org/project/{pkg}/\"])\n",
    "    return pd.DataFrame(\n",
    "        info_list,\n",
    "        columns=[\n",
    "            \"name\",\n",
    "            \"project_urls\",\n",
    "        ],\n",
    "    )\n",
    "\n",
    "\n",
    "def plot_communities(framework: str, package_names: list):\n",
    "    comm_nodes = [\n",
    "        line.strip(\"\\n\").split(\", \")\n",
    "        for line in open(f\"data/{framework}/{framework}_communities\")\n",
    "    ]\n",
    "    print(len(comm_nodes))\n",
    "\n",
    "    df = pd.DataFrame()\n",
    "    for i, comm in enumerate(comm_nodes):\n",
    "        tmp_df = comm_pkg_info(comm)\n",
    "        tmp_df.loc[:, \"comm_id\"] = i\n",
    "        df = df.append(tmp_df)\n",
    "    df.to_csv(f\"data/{framework}/{framework}_community_info.csv\", index=False)\n",
    "\n",
    "    edges = []\n",
    "    with open(f\"data/{framework}/{framework}_edges.csv\") as f:\n",
    "        for line in f:\n",
    "            down_pkg, up_pkg = line.strip(\"\\n\").split(\",\")\n",
    "            has_layer1_pkg = (down_pkg in package_names) or (up_pkg in package_names)\n",
    "            if not has_layer1_pkg:\n",
    "                edges.append((up_pkg, down_pkg))\n",
    "    nx_dig = nx.DiGraph()\n",
    "    nx_dig.add_edges_from(edges, weight=1)\n",
    "\n",
    "    for i, comm in enumerate(comm_nodes):\n",
    "        subg = nx_dig.subgraph(comm)\n",
    "        ig_graph = ig.Graph.from_networkx(subg)\n",
    "        ig.plot(\n",
    "            ig_graph,\n",
    "            f\"figures/{framework}/community_{i:03d}_{comm[0]}.pdf\",\n",
    "            vertex_size=8,\n",
    "            edge_arrow_width=0.6,\n",
    "            edge_arrow_size=0.6,\n",
    "        )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "8d9bfe9a-d45e-40ef-b6f7-6f4cc6e491b9",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-09-02T04:22:31.458226Z",
     "iopub.status.busy": "2022-09-02T04:22:31.458100Z",
     "iopub.status.idle": "2022-09-02T04:22:32.387255Z",
     "shell.execute_reply": "2022-09-02T04:22:32.386579Z",
     "shell.execute_reply.started": "2022-09-02T04:22:31.458211Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "131\n"
     ]
    }
   ],
   "source": [
    "plot_communities(\"TensorFlow\", [\"tensorflow\", \"tensorflow-cpu\", \"tensorflow-gpu\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "1543f8a4-9281-4969-8b53-80d5d1785825",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-09-02T04:22:32.388323Z",
     "iopub.status.busy": "2022-09-02T04:22:32.388072Z",
     "iopub.status.idle": "2022-09-02T04:22:33.462797Z",
     "shell.execute_reply": "2022-09-02T04:22:33.462175Z",
     "shell.execute_reply.started": "2022-09-02T04:22:32.388303Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "100\n"
     ]
    }
   ],
   "source": [
    "plot_communities(\"PyTorch\", [\"torch\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "b4cd2053-0054-4e79-b368-dca9b333b271",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-02-02T06:27:40.985007Z",
     "iopub.status.busy": "2023-02-02T06:27:40.984440Z",
     "iopub.status.idle": "2023-02-02T06:27:41.155505Z",
     "shell.execute_reply": "2023-02-02T06:27:41.154931Z",
     "shell.execute_reply.started": "2023-02-02T06:27:40.984955Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Representative</th>\n",
       "      <th>Size</th>\n",
       "      <th>Inspector1</th>\n",
       "      <th>Inspector2</th>\n",
       "      <th>Final</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>torchvision</td>\n",
       "      <td>657</td>\n",
       "      <td>Tree</td>\n",
       "      <td>Tree</td>\n",
       "      <td>Tree</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>pytorch-lightning</td>\n",
       "      <td>192</td>\n",
       "      <td>Forest</td>\n",
       "      <td>Tree</td>\n",
       "      <td>Forest</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>fastai</td>\n",
       "      <td>98</td>\n",
       "      <td>Forest</td>\n",
       "      <td>Forest</td>\n",
       "      <td>Forest</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>torchaudio</td>\n",
       "      <td>96</td>\n",
       "      <td>Forest</td>\n",
       "      <td>Forest</td>\n",
       "      <td>Forest</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>torchtext</td>\n",
       "      <td>94</td>\n",
       "      <td>Forest</td>\n",
       "      <td>Forest</td>\n",
       "      <td>Forest</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      Representative  Size Inspector1 Inspector2   Final\n",
       "0        torchvision   657       Tree       Tree    Tree\n",
       "1  pytorch-lightning   192     Forest       Tree  Forest\n",
       "2             fastai    98     Forest     Forest  Forest\n",
       "3         torchaudio    96     Forest     Forest  Forest\n",
       "4          torchtext    94     Forest     Forest  Forest"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "tf_shapes = pd.read_excel(\"data/community types.xlsx\", sheet_name=\"TensorFlow\")\n",
    "pt_shapes = pd.read_excel(\"data/community types.xlsx\", sheet_name=\"PyTorch\")\n",
    "pt_shapes.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "83d85dec-2743-4ef6-b474-24580287b86b",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-02-02T06:27:45.327405Z",
     "iopub.status.busy": "2023-02-02T06:27:45.326960Z",
     "iopub.status.idle": "2023-02-02T06:27:45.385712Z",
     "shell.execute_reply": "2023-02-02T06:27:45.385088Z",
     "shell.execute_reply.started": "2023-02-02T06:27:45.327367Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.9072512647554807, 0.895543545672012, 0.9183117767188563)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def calculate_cohen(rater1, rater2, categories):\n",
    "    df = pd.DataFrame({\"rater1\": rater1, \"rater2\": rater2})\n",
    "    df = (\n",
    "        df.groupby([\"rater1\", \"rater2\"])\n",
    "        .size()\n",
    "        .reset_index(name=\"count\")\n",
    "        .pivot_table(index=\"rater1\", columns=\"rater2\", values=\"count\", fill_value=0)\n",
    "    )\n",
    "    df = df[categories]\n",
    "    df = df.reindex(categories)\n",
    "\n",
    "    total_items = df.values.sum()\n",
    "    sum1 = df.sum(axis=1)\n",
    "    sum2 = df.sum(axis=0)\n",
    "\n",
    "    p_e = sum(sum1 * sum2) / total_items**2\n",
    "    p_0 = sum([df.loc[s, s] for s in categories]) / total_items\n",
    "\n",
    "    return (p_0 - p_e) / (1 - p_e)\n",
    "\n",
    "\n",
    "all_shapes = tf_shapes.append(pt_shapes, ignore_index=True)\n",
    "\n",
    "calculate_cohen(\n",
    "    all_shapes[\"Inspector1\"],\n",
    "    all_shapes[\"Inspector2\"],\n",
    "    all_shapes[\"Final\"].unique(),\n",
    "), calculate_cohen(\n",
    "    tf_shapes[\"Inspector1\"],\n",
    "    tf_shapes[\"Inspector2\"],\n",
    "    tf_shapes[\"Final\"].unique(),\n",
    "), calculate_cohen(\n",
    "    pt_shapes[\"Inspector1\"],\n",
    "    pt_shapes[\"Inspector2\"],\n",
    "    pt_shapes[\"Final\"].unique(),\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "cee29b2e-802f-4347-9109-ecd8166fd22c",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-02-02T06:27:48.765608Z",
     "iopub.status.busy": "2023-02-02T06:27:48.765101Z",
     "iopub.status.idle": "2023-02-02T06:27:48.774480Z",
     "shell.execute_reply": "2023-02-02T06:27:48.773517Z",
     "shell.execute_reply.started": "2023-02-02T06:27:48.765560Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "tf_shapes = tf_shapes[[\"Representative\", \"Size\", \"Final\"]].copy()\n",
    "pt_shapes = pt_shapes[[\"Representative\", \"Size\", \"Final\"]].copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "2f993191-eb66-457d-a0bb-e85120dc1fc3",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-02-02T06:27:51.229671Z",
     "iopub.status.busy": "2023-02-02T06:27:51.229143Z",
     "iopub.status.idle": "2023-02-02T06:27:51.244106Z",
     "shell.execute_reply": "2023-02-02T06:27:51.243536Z",
     "shell.execute_reply.started": "2023-02-02T06:27:51.229622Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([68, 26, 17, 13,  7]),\n",
       " Final\n",
       " Arrow     38\n",
       " Star      17\n",
       " Tree      20\n",
       " Forest    23\n",
       " Other      2\n",
       " dtype: int64)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tf_shapes.groupby(\"Final\").size().loc[[\"Arrow\", \"Star\", \"Tree\", \"Forest\", \"Other\"]].values, pt_shapes.groupby(\"Final\").size().loc[[\"Arrow\", \"Star\", \"Tree\", \"Forest\", \"Other\"]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "8e5c9340-70a1-4f15-930c-29b8727b46b8",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-02-02T06:27:54.266635Z",
     "iopub.status.busy": "2023-02-02T06:27:54.266115Z",
     "iopub.status.idle": "2023-02-02T06:27:55.538753Z",
     "shell.execute_reply": "2023-02-02T06:27:55.538294Z",
     "shell.execute_reply.started": "2023-02-02T06:27:54.266588Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 540x252 with 2 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 244,
       "width": 524
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import seaborn as sns\n",
    "\n",
    "%matplotlib inline\n",
    "%config InlineBackend.figure_formats=\"retina\"\n",
    "\n",
    "def plot_shape_dist():\n",
    "    labels = [\"Arrow\", \"Star\", \"Tree\", \"Forest\", \"Other\"]\n",
    "    tf_data = tf_shapes.groupby(\"Final\").size().loc[labels].values\n",
    "    pt_data = pt_shapes.groupby(\"Final\").size().loc[labels].values\n",
    "    \n",
    "    fig, axes = plt.subplots(1, 2, figsize=(7.5, 3.5))\n",
    "    colors = sns.color_palette(\"Set2\")\n",
    "\n",
    "    axes[0].pie(\n",
    "        tf_data, \n",
    "        colors=colors, \n",
    "        labels=labels, \n",
    "        autopct='%1.1f%%', \n",
    "        startangle=-180, \n",
    "        textprops={'fontsize':10}, \n",
    "        pctdistance=0.75, \n",
    "        counterclock=False,\n",
    "        wedgeprops=dict(edgecolor='w', linewidth=0.75)\n",
    "    )\n",
    "    axes[1].pie(\n",
    "        pt_data, \n",
    "        colors=colors, \n",
    "        labels=labels, \n",
    "        autopct='%1.1f%%', \n",
    "        startangle=180, \n",
    "        textprops={'fontsize':10}, \n",
    "        pctdistance=0.75, \n",
    "        counterclock=False,\n",
    "        wedgeprops=dict(edgecolor='w', linewidth=0.75)\n",
    "    )\n",
    "    plt.tight_layout() \n",
    "    fig.savefig(\"figures/shape_distribution.pdf\", bbox_inches=\"tight\")\n",
    "    \n",
    "    return\n",
    "\n",
    "plot_shape_dist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "0450131e-b3e9-47d0-b0bf-80ad6bf085af",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-09-02T04:22:33.699618Z",
     "iopub.status.busy": "2022-09-02T04:22:33.699438Z",
     "iopub.status.idle": "2022-09-02T04:22:33.705714Z",
     "shell.execute_reply": "2022-09-02T04:22:33.705219Z",
     "shell.execute_reply.started": "2022-09-02T04:22:33.699602Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(Final\n",
       " Arrow     136\n",
       " Forest    392\n",
       " Other      17\n",
       " Star      108\n",
       " Tree      239\n",
       " Name: Size, dtype: int64,\n",
       " Final\n",
       " Arrow       76\n",
       " Forest    1111\n",
       " Other        2\n",
       " Star        67\n",
       " Tree       788\n",
       " Name: Size, dtype: int64)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tf_shapes.groupby(\"Final\")[\"Size\"].sum(), pt_shapes.groupby(\"Final\")[\"Size\"].sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "dfa01442-9253-4248-85e1-c33d8735ca89",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-02-02T06:28:08.005804Z",
     "iopub.status.busy": "2023-02-02T06:28:08.005260Z",
     "iopub.status.idle": "2023-02-02T06:28:08.462762Z",
     "shell.execute_reply": "2023-02-02T06:28:08.462138Z",
     "shell.execute_reply.started": "2023-02-02T06:28:08.005754Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import networkx as nx\n",
    "\n",
    "def get_clusters():\n",
    "    tf_clusters = {}\n",
    "    pt_clusters = {}\n",
    "    with open(\"data/TensorFlow/TensorFlow_communities\") as fin:\n",
    "        for line in fin:\n",
    "            edges = []\n",
    "            items = line.strip(\"\\n\").split(\", \")\n",
    "            for row in tf_edges[\n",
    "                tf_edges[\"name\"].isin(items) & tf_edges[\"dependency\"].isin(items)\n",
    "            ].itertuples(index=False):\n",
    "                edges.append((row[1], row[0]))\n",
    "            G = nx.DiGraph(edges)\n",
    "            tf_clusters[items[0]] = G\n",
    "\n",
    "    with open(\"data/PyTorch/PyTorch_communities\") as fin:\n",
    "        for line in fin:\n",
    "            edges = []\n",
    "            items = line.strip(\"\\n\").split(\", \")\n",
    "            for row in pt_edges[\n",
    "                pt_edges[\"name\"].isin(items) & pt_edges[\"dependency\"].isin(items)\n",
    "            ].itertuples(index=False):\n",
    "                edges.append((row[1], row[0]))\n",
    "            G = nx.DiGraph(edges)\n",
    "            pt_clusters[items[0]] = G\n",
    "\n",
    "    return tf_clusters, pt_clusters\n",
    "\n",
    "\n",
    "tf_clusters, pt_clusters = get_clusters()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "c122dd94-7dc3-465c-94a2-2109863c6c93",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-02-02T07:30:55.243104Z",
     "iopub.status.busy": "2023-02-02T07:30:55.242377Z",
     "iopub.status.idle": "2023-02-02T07:30:55.591531Z",
     "shell.execute_reply": "2023-02-02T07:30:55.591028Z",
     "shell.execute_reply.started": "2023-02-02T07:30:55.243049Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 540x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 206,
       "width": 532
      },
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def tf_average_edge(s):\n",
    "    g = tf_clusters[s[\"Representative\"]]\n",
    "    return len(g.edges()) / len(g.nodes())\n",
    "\n",
    "\n",
    "def pt_average_edge(s):\n",
    "    g = pt_clusters[s[\"Representative\"]]\n",
    "    return len(g.edges()) / len(g.nodes())\n",
    "\n",
    "\n",
    "def cluster_centrality():\n",
    "\n",
    "    tf_shapes = pd.read_excel(\"data/community types.xlsx\", sheet_name=\"TensorFlow\")\n",
    "    pt_shapes = pd.read_excel(\"data/community types.xlsx\", sheet_name=\"PyTorch\")\n",
    "    tf_shapes = tf_shapes[[\"Representative\", \"Size\", \"Final\"]].copy()\n",
    "    pt_shapes = pt_shapes[[\"Representative\", \"Size\", \"Final\"]].copy()\n",
    "\n",
    "    tf_shapes[\"average edge\"] = tf_shapes.apply(tf_average_edge, axis=1)\n",
    "    pt_shapes[\"average edge\"] = pt_shapes.apply(pt_average_edge, axis=1)\n",
    "\n",
    "    shapes = [\"Arrow\", \"Star\", \"Tree\", \"Forest\"]\n",
    "    tf_data = (\n",
    "        tf_shapes[[\"Final\", \"average edge\"]].copy().rename(columns={\"Final\": \"Shape\"})\n",
    "    )\n",
    "    pt_data = (\n",
    "        pt_shapes[[\"Final\", \"average edge\"]].copy().rename(columns={\"Final\": \"Shape\"})\n",
    "    )\n",
    "    tf_data[\"Framework\"] = \"TensorFlow\"\n",
    "    pt_data[\"Framework\"] = \"PyTorch\"\n",
    "    data = pd.concat([tf_data, pt_data], ignore_index=True)\n",
    "    fig, ax = plt.subplots(1, 1, figsize=(7.5, 3))\n",
    "    palette = \"Accent\"\n",
    "    sns.boxplot(\n",
    "        x=\"Framework\",\n",
    "        y=\"average edge\",\n",
    "        data=data,\n",
    "        hue=\"Shape\",\n",
    "        hue_order=shapes,\n",
    "        order=[\"TensorFlow\", \"PyTorch\"],\n",
    "        ax=ax,\n",
    "        palette=palette,\n",
    "    )\n",
    "    ax.tick_params(axis=\"both\", which=\"major\", labelsize=12)\n",
    "    ax.set_xlabel(\"Framework\", fontsize=12)\n",
    "    ax.set_ylabel(\"Average Degree\", fontsize=12)\n",
    "    sns.move_legend(ax, loc=\"center\", ncol=2, fontsize=12, bbox_to_anchor=(0.65, 0.7))\n",
    "    ax.legend(loc=\"upper left\")\n",
    "    plt.tight_layout()\n",
    "    fig.savefig(\"figures/cluster_centralities.pdf\", bbox_inches=\"tight\")\n",
    "    return\n",
    "\n",
    "\n",
    "cluster_centrality()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "ba64581f-0142-4733-803f-3ab226dfae95",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-09-02T04:22:38.366646Z",
     "iopub.status.busy": "2022-09-02T04:22:38.366414Z",
     "iopub.status.idle": "2022-09-02T04:22:38.387092Z",
     "shell.execute_reply": "2022-09-02T04:22:38.386517Z",
     "shell.execute_reply.started": "2022-09-02T04:22:38.366628Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'tf:Arrow<Star': (6.5737977526055895e-21, True),\n",
       " 'tf:Star<Tree': (0.18772182434008544, False),\n",
       " 'tf:Tree<Forest': (0.008809046964535504, True),\n",
       " 'pt:Arrow<Star': (1.320724709469907e-12, True),\n",
       " 'pt:Star<Tree': (0.11225986379211311, False),\n",
       " 'pt:Tree<Forest': (0.00014025086628263644, True)}"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from multipy.fwer import holm_bonferroni\n",
    "from scipy.stats import mannwhitneyu\n",
    "\n",
    "\n",
    "def test_size(field):\n",
    "\n",
    "    ps = []\n",
    "    _, p1 = mannwhitneyu(\n",
    "        tf_shapes[tf_shapes[\"Final\"] == \"Arrow\"][field],\n",
    "        tf_shapes[tf_shapes[\"Final\"] == \"Star\"][field],\n",
    "        alternative=\"less\",\n",
    "    )\n",
    "    _, p2 = mannwhitneyu(\n",
    "        tf_shapes[tf_shapes[\"Final\"] == \"Star\"][field],\n",
    "        tf_shapes[tf_shapes[\"Final\"] == \"Tree\"][field],\n",
    "        alternative=\"less\",\n",
    "    )\n",
    "    _, p3 = mannwhitneyu(\n",
    "        tf_shapes[tf_shapes[\"Final\"] == \"Tree\"][field],\n",
    "        tf_shapes[tf_shapes[\"Final\"] == \"Forest\"][field],\n",
    "        alternative=\"less\",\n",
    "    )\n",
    "    _, p4 = mannwhitneyu(\n",
    "        pt_shapes[pt_shapes[\"Final\"] == \"Arrow\"][field],\n",
    "        pt_shapes[pt_shapes[\"Final\"] == \"Star\"][field],\n",
    "        alternative=\"less\",\n",
    "    )\n",
    "    _, p5 = mannwhitneyu(\n",
    "        pt_shapes[pt_shapes[\"Final\"] == \"Star\"][field],\n",
    "        pt_shapes[pt_shapes[\"Final\"] == \"Tree\"][field],\n",
    "        alternative=\"less\",\n",
    "    )\n",
    "    _, p6 = mannwhitneyu(\n",
    "        pt_shapes[pt_shapes[\"Final\"] == \"Tree\"][field],\n",
    "        pt_shapes[pt_shapes[\"Final\"] == \"Forest\"][field],\n",
    "        alternative=\"less\",\n",
    "    )\n",
    "\n",
    "    [a1, a2, a3, a4, a5, a6] = holm_bonferroni([p1, p2, p3, p4, p5, p6], alpha=0.05)\n",
    "    ps = {\n",
    "        \"tf:Arrow<Star\": (p1 * 6, a1),\n",
    "        \"tf:Star<Tree\": (p2 * 2, a2),\n",
    "        \"tf:Tree<Forest\": (p3 * 3, a3),\n",
    "        \"pt:Arrow<Star\": (p4 * 5, a4),\n",
    "        \"pt:Star<Tree\": (p5 * 1, a5),\n",
    "        \"pt:Tree<Forest\": (p6 * 4, a6),\n",
    "    }\n",
    "    return ps\n",
    "\n",
    "\n",
    "test_size(\"Size\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "8c14349e-816c-4133-9416-aa6c4e9a0839",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-09-02T04:22:38.387978Z",
     "iopub.status.busy": "2022-09-02T04:22:38.387788Z",
     "iopub.status.idle": "2022-09-02T04:22:38.439989Z",
     "shell.execute_reply": "2022-09-02T04:22:38.439445Z",
     "shell.execute_reply.started": "2022-09-02T04:22:38.387961Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'tf:Arrow<Star': (6.5737977526055895e-21, True),\n",
       " 'tf:Star<Tree': (0.0009716236951929591, True),\n",
       " 'tf:Tree<Forest': (0.0009901758186295724, True),\n",
       " 'pt:Arrow<Star': (1.320724709469907e-12, True),\n",
       " 'pt:Star<Tree': (2.1451227267292412e-06, True),\n",
       " 'pt:Tree<Forest': (0.0013643209467375651, True)}"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tf_shapes = pd.read_excel(\"data/community types.xlsx\", sheet_name=\"TensorFlow\")\n",
    "pt_shapes = pd.read_excel(\"data/community types.xlsx\", sheet_name=\"PyTorch\")\n",
    "tf_shapes = tf_shapes[[\"Representative\", \"Size\", \"Final\"]].copy()\n",
    "pt_shapes = pt_shapes[[\"Representative\", \"Size\", \"Final\"]].copy()\n",
    "\n",
    "tf_shapes[\"average edge\"] = tf_shapes.apply(tf_average_edge, axis=1)\n",
    "pt_shapes[\"average edge\"] = pt_shapes.apply(pt_average_edge, axis=1)\n",
    "test_size(\"average edge\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "d3beb54f-ef24-495c-8670-f85c7c934e20",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-09-02T04:22:38.440853Z",
     "iopub.status.busy": "2022-09-02T04:22:38.440668Z",
     "iopub.status.idle": "2022-09-02T04:22:38.447505Z",
     "shell.execute_reply": "2022-09-02T04:22:38.446951Z",
     "shell.execute_reply.started": "2022-09-02T04:22:38.440837Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(23, 20, 13, 11)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(pt_shapes[pt_shapes[\"Final\"] == \"Forest\"]), len(\n",
    "    pt_shapes[(pt_shapes[\"Final\"] == \"Forest\") & (pt_shapes[\"average edge\"] >= 1)]\n",
    "), len(tf_shapes[tf_shapes[\"Final\"] == \"Forest\"]), len(\n",
    "    tf_shapes[(tf_shapes[\"Final\"] == \"Forest\") & (tf_shapes[\"average edge\"] >= 1)]\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "b7973d19-5f8f-48cc-98d5-72738c795a81",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-09-02T04:22:38.448334Z",
     "iopub.status.busy": "2022-09-02T04:22:38.448156Z",
     "iopub.status.idle": "2022-09-02T04:22:38.454902Z",
     "shell.execute_reply": "2022-09-02T04:22:38.454311Z",
     "shell.execute_reply.started": "2022-09-02T04:22:38.448318Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(20, 15, 17, 9)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(pt_shapes[pt_shapes[\"Final\"] == \"Tree\"]), len(\n",
    "    pt_shapes[(pt_shapes[\"Final\"] == \"Tree\") & (pt_shapes[\"average edge\"] >= 1)]\n",
    "), len(tf_shapes[tf_shapes[\"Final\"] == \"Tree\"]), len(\n",
    "    tf_shapes[(tf_shapes[\"Final\"] == \"Tree\") & (tf_shapes[\"average edge\"] >= 1)]\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "ac82ff40-24e6-47f1-a4a7-3080ae98e014",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-09-02T04:22:38.455757Z",
     "iopub.status.busy": "2022-09-02T04:22:38.455579Z",
     "iopub.status.idle": "2022-09-02T04:22:38.461426Z",
     "shell.execute_reply": "2022-09-02T04:22:38.460883Z",
     "shell.execute_reply.started": "2022-09-02T04:22:38.455741Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Final\n",
       "Arrow     0.500000\n",
       "Forest    1.080000\n",
       "Other     1.000000\n",
       "Star      0.666667\n",
       "Tree      1.000000\n",
       "Name: average edge, dtype: float64"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pt_shapes.groupby(\"Final\")[\"average edge\"].median()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "7eaedca4-84bb-4fd2-bfba-39f5ea934454",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-09-02T04:22:38.462252Z",
     "iopub.status.busy": "2022-09-02T04:22:38.462075Z",
     "iopub.status.idle": "2022-09-02T04:22:38.467333Z",
     "shell.execute_reply": "2022-09-02T04:22:38.466881Z",
     "shell.execute_reply.started": "2022-09-02T04:22:38.462235Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Final\n",
       "Arrow     0.500000\n",
       "Forest    1.076923\n",
       "Other     0.833333\n",
       "Star      0.708333\n",
       "Tree      1.000000\n",
       "Name: average edge, dtype: float64"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tf_shapes.groupby(\"Final\")[\"average edge\"].median()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e1fe1ac0-5111-4c9c-a2f8-ffbfce7afb9c",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.8.13"
  },
  "vscode": {
   "interpreter": {
    "hash": "1e2195099b66390a9dc96fc724970e7f4fb2c16f6be654ca8121fdf641308df7"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}