more plots about summary of references

1 files changed, 70 insertions, 3 deletions

diff --git a/notebooks/cc/references.ipynb b/notebooks/cc/references.ipynb
index 28e589ec..bb79f585 100644
--- a/notebooks/cc/references.ipynb
+++ b/notebooks/cc/references.ipynb
@@ -342,18 +342,26 @@
    "outputs": [],
    "source": [
     "date_range = pd.date_range(df_id_rich.not_valid_before.min(), df_id_rich.not_valid_before.max())\n",
+    "dct_active_all = {}\n",
+    "dct_reference_rich_all = {}\n",
+    "dct_referenced_all = {}\n",
+    "dct_isolated_all = {}\n",
     "dct_active_others = {}\n",
     "dct_reference_rich_others = {}\n",
     "dct_active_smartcards = {}\n",
     "dct_reference_rich_smartcards = {}\n",
     "for date in tqdm(date_range):\n",
     "    active_certs = df_id_rich.loc[(date >= df_id_rich.not_valid_before) & (date <= df_id_rich.not_valid_after)]\n",
+    "    dct_active_all[date] = active_certs.shape[0]\n",
+    "    dct_isolated_all[date] = active_certs.loc[(active_certs.n_references == 0) & (active_certs.n_referenced_by == 0)].shape[0]\n",
     "    dct_active_others[date] = active_certs.loc[\n",
     "        active_certs.category != \"ICs, Smart Cards and Smart Card-Related Devices and Systems\"\n",
     "    ].shape[0]\n",
     "    dct_active_smartcards[date] = active_certs.loc[\n",
     "        active_certs.category == \"ICs, Smart Cards and Smart Card-Related Devices and Systems\"\n",
     "    ].shape[0]\n",
+    "    dct_reference_rich_all[date] = active_certs.loc[active_certs.n_references > 0].shape[0]\n",
+    "    dct_referenced_all[date] = active_certs.loc[active_certs.n_referenced_by > 0].shape[0]\n",
     "    dct_reference_rich_others[date] = active_certs.loc[\n",
     "        (active_certs.category != \"ICs, Smart Cards and Smart Card-Related Devices and Systems\")\n",
     "        & (active_certs.n_references > 0)\n",
@@ -376,11 +384,70 @@
     "df_active_vs_ref_rich.index.name = \"date\"\n",
     "df_active_vs_ref_rich = df_active_vs_ref_rich.reset_index()\n",
     "\n",
-    "\n",
-    "df_active_vs_ref_rich_melted = df_active_vs_ref_rich.melt(id_vars=[\"date\"], var_name=\"category\", value_name=\"number of certificates\")\n",
+    "df_active_vs_ref_rich_melted = df_active_vs_ref_rich.melt(\n",
+    "    id_vars=[\"date\"], var_name=\"category\", value_name=\"number of certificates\"\n",
+    ")\n",
     "g = sns.lineplot(data=df_active_vs_ref_rich_melted, x=\"date\", y=\"number of certificates\", hue=\"category\")\n",
-    "g.set(title=\"Number of active certificates vs. reference-rich certificates in time\", xlabel=\"Time\", ylabel=\"Number of certificates\")\n",
+    "g.set(\n",
+    "    title=\"Number of active certificates vs. reference-rich certificates in time\",\n",
+    "    xlabel=\"Time\",\n",
+    "    ylabel=\"Number of certificates\",\n",
+    ")\n",
     "plt.savefig(RESULTS_DIR / \"lienplot_n_active_certs_vs_n_references.pdf\", bbox_inches=\"tight\")\n",
+    "plt.show()\n",
+    "\n",
+    "\n",
+    "df_summary_references = pd.concat(\n",
+    "    [\n",
+    "        pd.Series(dct_active_all, name=\"active certificates\"),\n",
+    "        pd.Series(dct_reference_rich_all, name=\"ref. rich certificates\"),\n",
+    "        pd.Series(dct_referenced_all, name=\"referenced certificates\"),\n",
+    "        pd.Series(dct_isolated_all, name=\"isolated certificates\"),\n",
+    "    ], axis=1\n",
+    ")\n",
+    "df_summary_references.index.name = \"date\"\n",
+    "df_summary_references = df_summary_references.reset_index()\n",
+    "df_summary_references_melted = df_summary_references.melt(id_vars=[\"date\"], var_name=\"category\", value_name=\"number of certificates\")\n",
+    "\n",
+    "g = sns.lineplot(data=df_summary_references_melted, x=\"date\", y=\"number of certificates\", hue=\"category\", errorbar=None)\n",
+    "g.set(\n",
+    "    title=\"Number of active certificates vs. reference-rich vs. referenced certificates in time\",\n",
+    "    xlabel=\"Time\",\n",
+    "    ylabel=\"Number of certificates\",\n",
+    ")\n",
+    "plt.savefig(RESULTS_DIR / \"lineplot_references_summary.pdf\", bbox_inches=\"tight\")\n",
+    "plt.show()\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "df_summary_references"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "df_ratios = df_summary_references.copy()\n",
+    "df_ratios[\"ref. rich certificates\"] = df_ratios[\"ref. rich certificates\"] / df_ratios[\"active certificates\"]\n",
+    "df_ratios[\"referenced certificates\"] = df_ratios[\"referenced certificates\"] / df_ratios[\"active certificates\"]\n",
+    "df_ratios[\"isolated certificates\"] = df_ratios[\"isolated certificates\"] / df_ratios[\"active certificates\"]\n",
+    "df_ratios = df_ratios.drop(columns=[\"active certificates\"])\n",
+    "df_ratios_melted = df_ratios.melt(id_vars=[\"date\"], var_name=\"category\", value_name=\"ratio of certificates\")\n",
+    "\n",
+    "g = sns.lineplot(data=df_ratios_melted, x=\"date\", y=\"ratio of certificates\", hue=\"category\", errorbar=None)\n",
+    "g.set(\n",
+    "    title=\"ratio of reference-rich vs. referenced vs. isolated certificates in time\",\n",
+    "    xlabel=\"Time\",\n",
+    "    ylabel=\"Number of certificates\",\n",
+    ")\n",
+    "plt.savefig(RESULTS_DIR / \"lineplot_reference_ratio.pdf\", bbox_inches=\"tight\")\n",
     "plt.show()"
    ]
   },