X-Git-Url: https://us-delaporte.pagekite.me/gitweb/dotfiles.git/blobdiff_plain/5e378f911e9d78734849e8bb51487e82de876245..2d0d87d37d5fa18c22a74f80bb8f90b2dd882b01:/data/issues.ipynb

diff --git a/data/issues.ipynb b/data/issues.ipynb
index 6a68c2a..08d596b 100644
--- a/data/issues.ipynb
+++ b/data/issues.ipynb
@@ -7,20 +7,24 @@
    "outputs": [],
    "source": [
     "import platform\n",
+    "import matplotlib.pyplot as plt\n",
     "print(platform.python_version())"
    ]
   },
   {
-   "cell_type": "code",
-   "execution_count": null,
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": []
+  },
+  {
+   "cell_type": "markdown",
    "metadata": {},
-   "outputs": [],
    "source": [
-    "# Get ye data\n",
-    "import glob\n",
-    "import os\n",
+    "# Todo List\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": [
+    "# Get ye data\n",
+    "import glob\n",
     "import os\n",
     "\n",
+    "data_files = glob.glob(os.path.expanduser('~\\\\Data\\\\*.json'))\n",
+    "\n",
     "for file in data_files:\n",
-    "\tprint(f\"{file}\", os.path.getsize(file))\n"
+    "\tprint(f\"{file}\", os.path.getsize(file))\n",
+    "\n"
    ]
   },
   {
@@ -43,13 +52,14 @@
    "source": [
     "import pandas as pd\n",
     "\n",
-    "data_sets = []\n",
+    "data_frames = []\n",
+    "for file in data_files:\n",
     "\n",
-    "file = data_files[6]\n",
-    "print(file)\n",
+    "\tf = open(file, encoding='utf-8')\n",
+    "\tdf_temp = pd.read_json(f)\n",
+    "\tdata_frames.append(df_temp)\n",
     "\n",
-    "f = open(file, encoding='utf-8')\n",
-    "df = pd.read_json(f)"
+    "df = pd.concat(data_frames, ignore_index=True)"
    ]
   },
   {
@@ -63,15 +73,36 @@
     "\n"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "#### Pie Chart"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,
    "metadata": {},
    "outputs": [],
    "source": [
+    "\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)"
    ]
   },
   {
@@ -98,10 +129,9 @@
    "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()"
+    "\n",
+    "\n",
+    "print(mile_counts)"
    ]
   },
   {
@@ -110,9 +140,11 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "closed_months = pd.to_datetime(df['closedAt'], format='ISO8601')\n",
-    "df['closed_month'] = closed_months.dt.month\n",
-    "df.drop(columns=['closedAt'], inplace=True)"
+    "\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()"
    ]
   },
   {
@@ -121,7 +153,14 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "df['mile'] = [d.get('title') if d else None for d in df.milestone]"
+    "type(mile_counts)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "#### Plot of closed issues by month and milestone for one project\n"
    ]
   },
   {
@@ -130,9 +169,9 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "\n",
-    "dfg = df.groupby(['mile', 'closed_month']).count().reset_index()\n",
-    "dfg.drop(columns=['milestone', 'title', 'url'], inplace=True)\n"
+    "closed_months = pd.to_datetime(df['closedAt'], format='ISO8601')\n",
+    "df['closed_month'] = closed_months.dt.month\n",
+    "df.drop(columns=['closedAt'], inplace=True)"
    ]
   },
   {
@@ -141,15 +180,18 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "\n",
-    "print(dfg)"
+    "df['mile'] = [d.get('title') if d else None for d in df.milestone]"
    ]
   },
   {
-   "cell_type": "markdown",
+   "cell_type": "code",
+   "execution_count": null,
    "metadata": {},
+   "outputs": [],
    "source": [
-    "#### Plot of closed issues by month and milestone for one project\n"
+    "\n",
+    "dfg = df.groupby(['mile', 'closed_month']).count().reset_index()\n",
+    "dfg.drop(columns=['milestone', 'title', 'url'], inplace=True)\n"
    ]
   },
   {