update make figures

Author: Sergey Feldman

make_figures.ipynb

@@ -2,7 +2,7 @@
  "cells": [
   {
    "cell_type": "code",
-   "execution_count": 1,
+   "execution_count": 29,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -18,7 +18,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 4,
+   "execution_count": 30,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -43,7 +43,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
+   "execution_count": 31,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -65,7 +65,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 6,
+   "execution_count": 32,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -81,15 +81,35 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 7,
+   "execution_count": 39,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
       "Number of datasets each algorithm does best on:\n",
-      "Counter({'AutoGluon (sec=120)': 84, 'AutoGluon (sec=60)': 74, 'LightGBM (n_iter=25)': 74, 'LightGBM (n_iter=10)': 68, 'Logistic Regression': 64, 'Random Forest': 64, 'SVC': 35})\n"
+      "Counter({'AutoGluon (sec=120)': 84, 'AutoGluon (sec=60)': 74, 'LightGBM (n_iter=25)': 74, 'LightGBM (n_iter=10)': 68, 'Logistic Regression': 64, 'Random Forest': 64, 'SVC': 35}) \n",
+      "\n",
+      "Average performance for each algorithm: model\n",
+      "AutoGluon (sec=120)     0.887491\n",
+      "AutoGluon (sec=60)      0.886326\n",
+      "LightGBM (n_iter=10)    0.886359\n",
+      "LightGBM (n_iter=25)    0.888885\n",
+      "Logistic Regression     0.857380\n",
+      "Random Forest           0.886253\n",
+      "SVC                     0.852368\n",
+      "Name: mean_auroc, dtype: float64 \n",
+      "\n",
+      "Median performance for each algorithm: model\n",
+      "AutoGluon (sec=120)     0.924359\n",
+      "AutoGluon (sec=60)      0.925754\n",
+      "LightGBM (n_iter=10)    0.924920\n",
+      "LightGBM (n_iter=25)    0.927121\n",
+      "Logistic Regression     0.891617\n",
+      "Random Forest           0.922966\n",
+      "SVC                     0.880665\n",
+      "Name: mean_auroc, dtype: float64\n"
      ]
     }
    ],
@@ -103,12 +123,14 @@
     "    winning_algorithms.extend(df_sub.loc[df_sub['mean_auroc'] >= highest_auroc * 0.995, 'model'])\n",
     "\n",
     "print('Number of datasets each algorithm does best on:')\n",
-    "print(Counter(winning_algorithms))"
+    "print(Counter(winning_algorithms), '\\n')\n",
+    "print('Average performance for each algorithm:', results_df.groupby('model')['mean_auroc'].mean(), '\\n')\n",
+    "print('Median performance for each algorithm:', results_df.groupby('model')['mean_auroc'].median())"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 27,
+   "execution_count": 35,
    "metadata": {},
    "outputs": [
     {
@@ -117,7 +139,7 @@
        "<AxesSubplot:xlabel='mean_auroc', ylabel='model'>"
       ]
      },
-     "execution_count": 27,
+     "execution_count": 35,
      "metadata": {},
      "output_type": "execute_result"
     },