Add a pie chart

diff --git a/data/issues.ipynb b/data/issues.ipynb
index 011beec2ea3144c44dfd250c94619c3e28411f66..6a68c2a0633103d850adbd7563e40b552aff4a6a 100644 (file)
--- a/data/issues.ipynb
+++ b/data/issues.ipynb
@@ -45,7 +45,7 @@
     "\n",
     "data_sets = []\n",
     "\n",
-    "file = data_files[2]\n",
+    "file = data_files[6]\n",
     "print(file)\n",
     "\n",
     "f = open(file, encoding='utf-8')\n",
@@ -63,6 +63,47 @@
     "\n"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "dfp = df\n",
+    "dfp['mile'] = [d.get('title') if d else None for d in dfp.milestone]\n",
+    "dfp.drop(columns=['closedAt','milestone', 'url'], inplace=True)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "\n",
+    "mile_counts = dfp.groupby('mile')['title'].nunique()\n",
+    "dfp['count'] = dfp['mile'].map(mile_counts)\n",
+    "# dfp.drop(columns=['title'], inplace=True)\n",
+    "dfp['mile'] = df['mile'].drop_duplicates()\n",
+    "dfp.dropna(inplace=True)\n",
+    "dfp = dfp[dfp['count'] > 5]\n",
+    "print(dfp)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "mile_counts = df.groupby('mile')['count'].sum()\n",
+    "mile_counts = mile_counts.loc[lambda x: x > 5]\n",
+    "plt.pie(mile_counts, labels=mile_counts.index, autopct='%1.1f%%')\n",
+    "plt.title(f\"Tasks by Milestone\")\n",
+    "plt.axis('equal')\n",
+    "plt.show()"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,
@@ -113,7 +154,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 32,
+   "execution_count": null,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -142,7 +183,8 @@
    "outputs": [],
    "source": [
     "plt.figure(figsize=(10, 6))\n",
-    "sns.barplot(x='closed_month', y='count', hue='mile', data=dfg)\n",
+    "sns.pie(x='closed_month', y='count', hue='mile', data=dfg)\n",
+    "plt.pie(dfg, )\n",
     "plt.xlabel('Closed Month')\n",
     "plt.ylabel('ClosedAt')\n",
     "plt.title('Stacked Line Chart: Milestone Effort by Issue Close Date')\n",