Add pie chart for combination of all data files

[dotfiles.git] / data / issues.ipynb

diff --git a/data/issues.ipynb b/data/issues.ipynb
index 7608fd74725e3d5037de9943f4c5765fdc252ef7..b65e1d7ca7d2645bdf35ceea09191beef3e8ab83 100644 (file)
--- a/data/issues.ipynb
+++ b/data/issues.ipynb
@@ -43,13 +43,14 @@
    "source": [
     "import pandas as pd\n",
     "\n",
    "source": [
     "import pandas as pd\n",
     "\n",

-    "data_sets = []\n",
+    "data_frames = []\n",
+    "for file in data_files:\n",

     "\n",
     "\n",

-    "file = data_files[2]\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",
     "\n",

-    "f = open(file, encoding='utf-8')\n",
-    "df = pd.read_json(f)"
+    "df = pd.concat(data_frames, ignore_index=True)"

    ]
   },
   {
    ]
   },
   {


@@ -63,15 +64,22 @@
     "\n"
    ]
   },
     "\n"
    ]
   },

+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "#### Pie Chart"
+   ]
+  },

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

    ]
   },
   {
    ]
   },
   {


@@ -80,7 +88,14 @@
    "metadata": {},
    "outputs": [],
    "source": [
    "metadata": {},
    "outputs": [],
    "source": [

-    "df['mile'] = [d.get('title') if d else None for d in df.milestone]"
+    "\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)"

    ]
   },
   {
    ]
   },
   {


@@ -89,9 +104,39 @@
    "metadata": {},
    "outputs": [],
    "source": [
    "metadata": {},
    "outputs": [],
    "source": [

-    "\n",
-    "grp_df = df.groupby(['mile', 'closed_month']).count().reset_index()\n",
-    "grp_df.drop(columns=['milestone', 'title', 'url'], inplace=True)"
+    "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": "markdown",
+   "metadata": {},
+   "source": [
+    "#### Plot of closed issues by month and milestone for one project\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "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)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "df['mile'] = [d.get('title') if d else None for d in df.milestone]"

    ]
   },
   {
    ]
   },
   {


@@ -101,14 +146,21 @@
    "outputs": [],
    "source": [
     "\n",
    "outputs": [],
    "source": [
     "\n",

-    "print(grp_df)"
+    "dfg = df.groupby(['mile', 'closed_month']).count().reset_index()\n",
+    "dfg.drop(columns=['milestone', 'title', 'url'], inplace=True)\n"

    ]
   },
   {
    ]
   },
   {

-   "cell_type": "markdown",
+   "cell_type": "code",
+   "execution_count": null,

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

+   "outputs": [],

    "source": [
    "source": [

-    "#### Plot of closed issues by month and milestone for one project\n"
+    "exclude_from_chart = [\n",
+    "\t'Job Aid Ideas']\n",
+    "\n",
+    "for ex in exclude_from_chart:\n",
+    "\tdfg = dfg[dfg['mile'] != ex]\n"

    ]
   },
   {
    ]
   },
   {


@@ -129,7 +181,8 @@
    "outputs": [],
    "source": [
     "plt.figure(figsize=(10, 6))\n",
    "outputs": [],
    "source": [
     "plt.figure(figsize=(10, 6))\n",

-    "sns.barplot(x='closed_month', y='url', hue='mile', data=grp_df)\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",
     "plt.xlabel('Closed Month')\n",
     "plt.ylabel('ClosedAt')\n",
     "plt.title('Stacked Line Chart: Milestone Effort by Issue Close Date')\n",