Add year metadata column

[dotfiles.git] / data / issues.ipynb

diff --git a/data/issues.ipynb b/data/issues.ipynb
index b65e1d7ca7d2645bdf35ceea09191beef3e8ab83..08d596b1b3916d78e94e01406fa07061b98c190d 100644 (file)
--- a/data/issues.ipynb
+++ b/data/issues.ipynb
@@ -7,20 +7,24 @@
    "outputs": [],
    "source": [
     "import platform\n",
    "outputs": [],
    "source": [
     "import platform\n",

+    "import matplotlib.pyplot as plt\n",

     "print(platform.python_version())"
    ]
   },
   {
     "print(platform.python_version())"
    ]
   },
   {

-   "cell_type": "code",
-   "execution_count": null,
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": []
+  },
+  {
+   "cell_type": "markdown",

    "metadata": {},
    "metadata": {},

-   "outputs": [],

    "source": [
    "source": [

-    "# Get ye data\n",
-    "import glob\n",
-    "import os\n",
+    "# Todo List\n",

     "\n",
     "\n",

-    "data_files = glob.glob(os.path.expanduser('~\\\\Data\\\\*.json'))\n"
+    "- [x] add calculated year to the metadata\n",
+    "- [ ] make one chart per year\n",
+    "- [x] make sure all the data files are getting loaded"

    ]
   },
   {
    ]
   },
   {


@@ -29,10 +33,15 @@
    "metadata": {},
    "outputs": [],
    "source": [
    "metadata": {},
    "outputs": [],
    "source": [

+    "# Get ye data\n",
+    "import glob\n",

     "import os\n",
     "\n",
     "import os\n",
     "\n",

+    "data_files = glob.glob(os.path.expanduser('~\\\\Data\\\\*.json'))\n",
+    "\n",

     "for file in data_files:\n",
     "for file in data_files:\n",

-    "\tprint(f\"{file}\", os.path.getsize(file))\n"
+    "\tprint(f\"{file}\", os.path.getsize(file))\n",
+    "\n"

    ]
   },
   {
    ]
   },
   {


@@ -77,9 +86,23 @@
    "metadata": {},
    "outputs": [],
    "source": [
    "metadata": {},
    "outputs": [],
    "source": [

+    "\n",

     "dfp = df\n",
     "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)"
+    "# dfp['mile'] = [m.get('title') if m else None for m in dfp.milestone]\n",
+    "dfp['mile'] = [m.get('title') if hasattr(m, 'get') else None for m in dfp['milestone']]\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "closed_dates = pd.to_datetime(df['closedAt'], format='ISO8601')\n",
+    "dfp['year'] = closed_dates.dt.year\n",
+    "\n",
+    "dfp.drop(columns=['closedAt','milestone', 'url'], inplace=True)\n",
+    "print(dfp)"

    ]
   },
   {
    ]
   },
   {


@@ -106,12 +129,33 @@
    "source": [
     "mile_counts = df.groupby('mile')['count'].sum()\n",
     "mile_counts = mile_counts.loc[lambda x: x > 5]\n",
    "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",
+    "\n",
+    "\n",
+    "print(mile_counts)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "\n",
+    "plt.pie(mile_counts, labels=mile_counts.index, autopct='%0.1f%%')\n",

     "plt.title(f\"Tasks by Milestone\")\n",
     "plt.axis('equal')\n",
     "plt.show()"
    ]
   },
     "plt.title(f\"Tasks by Milestone\")\n",
     "plt.axis('equal')\n",
     "plt.show()"
    ]
   },

+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "type(mile_counts)"
+   ]
+  },

   {
    "cell_type": "markdown",
    "metadata": {},
   {
    "cell_type": "markdown",
    "metadata": {},