analysis: better growth and tag figures

analysis.ipynb

@@ -161,13 +161,18 @@
     "d['track_id'].plot(ax=ax1)\n",
     "d['indicator'].cumsum().plot(ax=ax1)\n",
     "ax1.set_ylabel('#tracks')\n",
-    "ax1.legend(['track_id', 'collected'], loc='lower right')\n",
+    "ax1.set_ylim(0, 160000)\n",
     "\n",
     "d['indicator'].plot(ax=ax2, style='r')  # needed for no apparent reason\n",
-    "color = (0.7686274509803922, 0.3058823529411765, 0.3215686274509804)\n",
+    "color = sns.color_palette('deep', 3)[2]\n",
     "d['indicator'].resample('2M').sum().fillna(0).plot(ax=ax2, style='--', color=color)\n",
     "ax2.set_ylabel('#tracks added')\n",
-    "ax2.set_ylim(900, 3800)\n",
+    "ax2.set_ylim(500, 4500)\n",
+    "ax2.grid(False)\n",
+    "\n",
+    "lns = ax1.get_lines() + [ax2.get_lines()[1]]\n",
+    "ax1.legend(lns, ['largest track id', '#tracks still present', '#tracks added per 2 months'], loc='lower right')\n",
+    "\n",
     "plt.savefig('growth.pdf')"
    ]
   },
@@ -410,6 +415,7 @@
     "d = tracks['track', 'tags'].apply(len)\n",
     "print(d.min(), d.max())\n",
     "d = d[(d.values >= 1) & (d.values <= 16)]\n",
+    "color = sns.color_palette('deep', 3)[2]\n",
     "sns.distplot(d, kde=False, bins=d.max()-d.min(), label='track', ax=ax1, color=color)\n",
     "\n",
     "d = tracks.drop_duplicates(('album', 'id'))\n",
@@ -428,8 +434,11 @@
     "ax1.set_ylabel('#tracks')\n",
     "ax2.set_ylabel('#artists   /   #albums')\n",
     "ax1.set_xlim(d.min(), d.max())\n",
-    "ax1.legend()\n",
-    "ax2.legend(loc='upper center')\n",
+    "ax1.set_ylim(0, 5000)\n",
+    "ax2.set_ylim(0, 500)\n",
+    "ax1.legend(loc='upper center')\n",
+    "ax2.legend(loc='upper right')\n",
+    "ax2.grid(False)\n",
     "\n",
     "plt.tight_layout()\n",
     "plt.savefig('tag_distribution.pdf')"