Fix the visualization

analysis.ipynb

@@ -5,9 +5,12 @@
    "id": "initial_id",
    "metadata": {
     "collapsed": true,
+    "jupyter": {
+     "outputs_hidden": true
+    },
     "ExecuteTime": {
-     "end_time": "2024-05-01T13:07:35.991719Z",
-     "start_time": "2024-05-01T13:07:16.672667Z"
+     "end_time": "2024-05-01T15:23:17.507403Z",
+     "start_time": "2024-05-01T15:23:17.497406Z"
     }
    },
    "source": [
@@ -15,41 +18,23 @@
     "\n",
     "import config"
    ],
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "D:\\petrtsv\\work\\jetbrains\\commit-rewriting-processing\\.venv\\lib\\site-packages\\tqdm\\auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
-      "  from .autonotebook import tqdm as notebook_tqdm\n",
-      "[nltk_data] Downloading package wordnet to C:\\Users\\Petr\n",
-      "[nltk_data]     Tsvetkov\\AppData\\Roaming\\nltk_data...\n",
-      "[nltk_data]   Package wordnet is already up-to-date!\n",
-      "[nltk_data] Downloading package punkt to C:\\Users\\Petr\n",
-      "[nltk_data]     Tsvetkov\\AppData\\Roaming\\nltk_data...\n",
-      "[nltk_data]   Package punkt is already up-to-date!\n",
-      "[nltk_data] Downloading package omw-1.4 to C:\\Users\\Petr\n",
-      "[nltk_data]     Tsvetkov\\AppData\\Roaming\\nltk_data...\n",
-      "[nltk_data]   Package omw-1.4 is already up-to-date!\n"
-     ]
-    }
-   ],
-   "execution_count": 20
+   "outputs": [],
+   "execution_count": 7
   },
   {
+   "cell_type": "code",
+   "id": "2ac8757a17e62293",
    "metadata": {
     "ExecuteTime": {
-     "end_time": "2024-05-01T12:57:08.596650Z",
-     "start_time": "2024-05-01T12:57:08.435650Z"
+     "end_time": "2024-05-01T15:23:19.365525Z",
+     "start_time": "2024-05-01T15:23:19.120308Z"
     }
    },
-   "cell_type": "code",
    "source": [
     "df = pd.read_csv(config.SYNTHETIC_DATASET_ARTIFACT, index_col=0)\n",
     "\n",
     "df.head()"
    ],
-   "id": "2ac8757a17e62293",
    "outputs": [
     {
      "data": {
@@ -316,6 +301,30 @@
        "</div>"
       ]
      },
+     "execution_count": 8,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "execution_count": 8
+  },
+  {
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2024-05-01T15:11:08.418257Z",
+     "start_time": "2024-05-01T15:11:08.408943Z"
+    }
+   },
+   "cell_type": "code",
+   "source": "len(set(df['session'].to_list()))",
+   "id": "4bcbc0f1d3d6d248",
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "9"
+      ]
+     },
      "execution_count": 6,
      "metadata": {},
      "output_type": "execute_result"
@@ -324,18 +333,15 @@
    "execution_count": 6
   },
   {
+   "cell_type": "code",
+   "execution_count": 15,
+   "id": "d19c12dd10b25c75",
    "metadata": {
     "ExecuteTime": {
      "end_time": "2024-05-01T13:02:40.761645Z",
      "start_time": "2024-05-01T13:02:40.740647Z"
     }
    },
-   "cell_type": "code",
-   "source": [
-    "rel_metrics = [col.split(\"_\")[0] for col in df.columns if col.endswith(\"_related\")]\n",
-    "rel_metrics"
-   ],
-   "id": "d19c12dd10b25c75",
    "outputs": [
     {
      "data": {
@@ -348,21 +354,21 @@
      "output_type": "execute_result"
     }
    ],
-   "execution_count": 15
+   "source": [
+    "rel_metrics = [col.split(\"_\")[0] for col in df.columns if col.endswith(\"_related\")]\n",
+    "rel_metrics"
+   ]
   },
   {
+   "cell_type": "code",
+   "execution_count": 16,
+   "id": "79d644cd780b28a1",
    "metadata": {
     "ExecuteTime": {
      "end_time": "2024-05-01T13:02:44.072037Z",
      "start_time": "2024-05-01T13:02:44.055039Z"
     }
    },
-   "cell_type": "code",
-   "source": [
-    "ind_metrics = [col.split(\"_\")[0] for col in df.columns if col.endswith(\"_independent\")]\n",
-    "ind_metrics"
-   ],
-   "id": "79d644cd780b28a1",
    "outputs": [
     {
      "data": {
@@ -385,96 +391,24 @@
      "output_type": "execute_result"
     }
    ],
-   "execution_count": 16
+   "source": [
+    "ind_metrics = [col.split(\"_\")[0] for col in df.columns if col.endswith(\"_independent\")]\n",
+    "ind_metrics"
+   ]
   },
   {
+   "cell_type": "code",
+   "execution_count": 19,
+   "id": "fdc5ae636bffbc8b",
    "metadata": {
     "ExecuteTime": {
      "end_time": "2024-05-01T13:03:52.623346Z",
      "start_time": "2024-05-01T13:03:52.577076Z"
     }
    },
-   "cell_type": "code",
-   "source": [
-    "AGGREGATION = {\"hash\": [\"count\"]}\n",
-    "\n",
-    "for metric in rel_metrics:\n",
-    "    AGGREGATION[f\"{metric}_related\"] = [\"mean\"]\n",
-    "\n",
-    "for metric in ind_metrics:\n",
-    "    AGGREGATION[f\"{metric}_independent\"] = [\"mean\"]\n",
-    "\n",
-    "df.groupby(by=[\"end_to_start\", \"start_to_end\"]).agg(AGGREGATION)"
-   ],
-   "id": "fdc5ae636bffbc8b",
    "outputs": [
     {
      "data": {
-      "text/plain": [
-       "                           hash editdist_related edittime_related  \\\n",
-       "                          count             mean             mean   \n",
-       "end_to_start start_to_end                                           \n",
-       "False        False           43       355.441860      364099.0625   \n",
-       "             True           129       406.627907              NaN   \n",
-       "True         False          129       433.899225              NaN   \n",
-       "             True           387       444.509044              NaN   \n",
-       "\n",
-       "                          gptscore-ref-1-req_independent  \\\n",
-       "                                                    mean   \n",
-       "end_to_start start_to_end                                  \n",
-       "False        False                              7.255814   \n",
-       "             True                               7.217054   \n",
-       "True         False                              7.356589   \n",
-       "             True                               7.312661   \n",
-       "\n",
-       "                          gptscore-noref-1-req_independent  \\\n",
-       "                                                      mean   \n",
-       "end_to_start start_to_end                                    \n",
-       "False        False                                8.116279   \n",
-       "             True                                 8.178295   \n",
-       "True         False                                8.302326   \n",
-       "             True                                 8.276486   \n",
-       "\n",
-       "                          editdist_independent bleu_independent  \\\n",
-       "                                          mean             mean   \n",
-       "end_to_start start_to_end                                         \n",
-       "False        False                  491.069767         0.012805   \n",
-       "             True                   491.069767         0.012805   \n",
-       "True         False                  534.015504         0.009542   \n",
-       "             True                   534.015504         0.009542   \n",
-       "\n",
-       "                          meteor_independent rouge1_independent  \\\n",
-       "                                        mean               mean   \n",
-       "end_to_start start_to_end                                         \n",
-       "False        False                  0.224961           0.202063   \n",
-       "             True                   0.224961           0.202063   \n",
-       "True         False                  0.221893           0.205151   \n",
-       "             True                   0.221893           0.205151   \n",
-       "\n",
-       "                          rouge2_independent rougeL_independent  \\\n",
-       "                                        mean               mean   \n",
-       "end_to_start start_to_end                                         \n",
-       "False        False                  0.040718           0.136427   \n",
-       "             True                   0.040718           0.136427   \n",
-       "True         False                  0.039033           0.134114   \n",
-       "             True                   0.039033           0.134114   \n",
-       "\n",
-       "                          bertscore_independent chrF_independent  \\\n",
-       "                                           mean             mean   \n",
-       "end_to_start start_to_end                                          \n",
-       "False        False                     0.780266        32.067005   \n",
-       "             True                      0.780266        32.067005   \n",
-       "True         False                     0.777162        31.753065   \n",
-       "             True                      0.777162        31.753065   \n",
-       "\n",
-       "                          ter_independent  \n",
-       "                                     mean  \n",
-       "end_to_start start_to_end                  \n",
-       "False        False             312.732989  \n",
-       "             True              312.732989  \n",
-       "True         False             317.717517  \n",
-       "             True              317.717517  "
-      ],
       "text/html": [
        "<div>\n",
        "<style scoped>\n",
@@ -625,6 +559,71 @@
        "  </tbody>\n",
        "</table>\n",
        "</div>"
+      ],
+      "text/plain": [
+       "                           hash editdist_related edittime_related  \\\n",
+       "                          count             mean             mean   \n",
+       "end_to_start start_to_end                                           \n",
+       "False        False           43       355.441860      364099.0625   \n",
+       "             True           129       406.627907              NaN   \n",
+       "True         False          129       433.899225              NaN   \n",
+       "             True           387       444.509044              NaN   \n",
+       "\n",
+       "                          gptscore-ref-1-req_independent  \\\n",
+       "                                                    mean   \n",
+       "end_to_start start_to_end                                  \n",
+       "False        False                              7.255814   \n",
+       "             True                               7.217054   \n",
+       "True         False                              7.356589   \n",
+       "             True                               7.312661   \n",
+       "\n",
+       "                          gptscore-noref-1-req_independent  \\\n",
+       "                                                      mean   \n",
+       "end_to_start start_to_end                                    \n",
+       "False        False                                8.116279   \n",
+       "             True                                 8.178295   \n",
+       "True         False                                8.302326   \n",
+       "             True                                 8.276486   \n",
+       "\n",
+       "                          editdist_independent bleu_independent  \\\n",
+       "                                          mean             mean   \n",
+       "end_to_start start_to_end                                         \n",
+       "False        False                  491.069767         0.012805   \n",
+       "             True                   491.069767         0.012805   \n",
+       "True         False                  534.015504         0.009542   \n",
+       "             True                   534.015504         0.009542   \n",
+       "\n",
+       "                          meteor_independent rouge1_independent  \\\n",
+       "                                        mean               mean   \n",
+       "end_to_start start_to_end                                         \n",
+       "False        False                  0.224961           0.202063   \n",
+       "             True                   0.224961           0.202063   \n",
+       "True         False                  0.221893           0.205151   \n",
+       "             True                   0.221893           0.205151   \n",
+       "\n",
+       "                          rouge2_independent rougeL_independent  \\\n",
+       "                                        mean               mean   \n",
+       "end_to_start start_to_end                                         \n",
+       "False        False                  0.040718           0.136427   \n",
+       "             True                   0.040718           0.136427   \n",
+       "True         False                  0.039033           0.134114   \n",
+       "             True                   0.039033           0.134114   \n",
+       "\n",
+       "                          bertscore_independent chrF_independent  \\\n",
+       "                                           mean             mean   \n",
+       "end_to_start start_to_end                                          \n",
+       "False        False                     0.780266        32.067005   \n",
+       "             True                      0.780266        32.067005   \n",
+       "True         False                     0.777162        31.753065   \n",
+       "             True                      0.777162        31.753065   \n",
+       "\n",
+       "                          ter_independent  \n",
+       "                                     mean  \n",
+       "end_to_start start_to_end                  \n",
+       "False        False             312.732989  \n",
+       "             True              312.732989  \n",
+       "True         False             317.717517  \n",
+       "             True              317.717517  "
       ]
      },
      "execution_count": 19,
@@ -632,100 +631,31 @@
      "output_type": "execute_result"
     }
    ],
-   "execution_count": 19
+   "source": [
+    "AGGREGATION = {\"hash\": [\"count\"]}\n",
+    "\n",
+    "for metric in rel_metrics:\n",
+    "    AGGREGATION[f\"{metric}_related\"] = [\"mean\"]\n",
+    "\n",
+    "for metric in ind_metrics:\n",
+    "    AGGREGATION[f\"{metric}_independent\"] = [\"mean\"]\n",
+    "\n",
+    "df.groupby(by=[\"end_to_start\", \"start_to_end\"]).agg(AGGREGATION)"
+   ]
   },
   {
+   "cell_type": "code",
+   "execution_count": 47,
+   "id": "3429b60eab154b79",
    "metadata": {
     "ExecuteTime": {
      "end_time": "2024-05-01T13:42:57.052768Z",
      "start_time": "2024-05-01T13:42:56.812556Z"
     }
    },
-   "cell_type": "code",
-   "source": "",
-   "id": "3429b60eab154b79",
    "outputs": [
     {
      "data": {
-      "text/plain": [
-       "                                    all              golden            \\\n",
-       "                               spearman   pearson  spearman   pearson   \n",
-       "relative independent                                                    \n",
-       "editdist bertscore            -0.184962 -0.129057 -0.316215 -0.254700   \n",
-       "         bleu                  0.260118  0.185995  0.269028  0.259690   \n",
-       "         chrF                 -0.199200 -0.129029 -0.343201 -0.300656   \n",
-       "         editdist              0.909934  0.910641  0.710772  0.662808   \n",
-       "         gptscore-noref-1-req  0.032048  0.055364  0.155510  0.048588   \n",
-       "         gptscore-ref-1-req    0.024550  0.035295 -0.009830 -0.062574   \n",
-       "         meteor                0.336016  0.371949  0.068034  0.173237   \n",
-       "         rouge1               -0.077574 -0.043738 -0.187349 -0.163230   \n",
-       "         rouge2                0.414256  0.340732  0.276139  0.332087   \n",
-       "         rougeL                0.006513 -0.008078 -0.041502 -0.034867   \n",
-       "         ter                   0.618095  0.385515  0.575614  0.501385   \n",
-       "edittime bertscore             0.140481  0.158807  0.140481  0.158807   \n",
-       "         bleu                  0.302380  0.326167  0.302380  0.326167   \n",
-       "         chrF                  0.079802  0.184202  0.079802  0.184202   \n",
-       "         editdist              0.252645  0.411131  0.252645  0.411131   \n",
-       "         gptscore-noref-1-req  0.206465  0.026235  0.206465  0.026235   \n",
-       "         gptscore-ref-1-req    0.130419 -0.055218  0.130419 -0.055218   \n",
-       "         meteor                0.253380  0.403564  0.253380  0.403564   \n",
-       "         rouge1                0.155926  0.136971  0.155926  0.136971   \n",
-       "         rouge2                0.218822  0.281944  0.218822  0.281944   \n",
-       "         rougeL                0.071344  0.091196  0.071344  0.091196   \n",
-       "         ter                   0.305601  0.062616  0.305601  0.062616   \n",
-       "\n",
-       "                                   +s2e                +e2s            \\\n",
-       "                               spearman   pearson  spearman   pearson   \n",
-       "relative independent                                                    \n",
-       "editdist bertscore            -0.308494 -0.113525 -0.181393 -0.165924   \n",
-       "         bleu                  0.512841  0.502827  0.109831  0.068138   \n",
-       "         chrF                 -0.238124 -0.064922 -0.233123 -0.201726   \n",
-       "         editdist              0.950494  0.935064  0.861930  0.878118   \n",
-       "         gptscore-noref-1-req  0.067857  0.047215 -0.029048 -0.013128   \n",
-       "         gptscore-ref-1-req   -0.015178 -0.036001  0.071345  0.087584   \n",
-       "         meteor                0.203616  0.425775  0.372598  0.360051   \n",
-       "         rouge1               -0.139874 -0.065543 -0.082093 -0.035603   \n",
-       "         rouge2                0.523559  0.537560  0.323911  0.282872   \n",
-       "         rougeL               -0.022288 -0.004664  0.012409  0.016372   \n",
-       "         ter                   0.774086  0.462554  0.529338  0.388592   \n",
-       "edittime bertscore                  NaN       NaN       NaN       NaN   \n",
-       "         bleu                       NaN       NaN       NaN       NaN   \n",
-       "         chrF                       NaN       NaN       NaN       NaN   \n",
-       "         editdist                   NaN       NaN       NaN       NaN   \n",
-       "         gptscore-noref-1-req       NaN       NaN       NaN       NaN   \n",
-       "         gptscore-ref-1-req         NaN       NaN       NaN       NaN   \n",
-       "         meteor                     NaN       NaN       NaN       NaN   \n",
-       "         rouge1                     NaN       NaN       NaN       NaN   \n",
-       "         rouge2                     NaN       NaN       NaN       NaN   \n",
-       "         rougeL                     NaN       NaN       NaN       NaN   \n",
-       "         ter                        NaN       NaN       NaN       NaN   \n",
-       "\n",
-       "                               +e2s+s2e            \n",
-       "                               spearman   pearson  \n",
-       "relative independent                               \n",
-       "editdist bertscore            -0.135421 -0.091748  \n",
-       "         bleu                  0.229712  0.145062  \n",
-       "         chrF                 -0.156914 -0.093376  \n",
-       "         editdist              0.939318  0.962305  \n",
-       "         gptscore-noref-1-req  0.012102  0.066882  \n",
-       "         gptscore-ref-1-req    0.013012  0.033618  \n",
-       "         meteor                0.392262  0.401802  \n",
-       "         rouge1               -0.054034 -0.030799  \n",
-       "         rouge2                0.433859  0.324538  \n",
-       "         rougeL                0.021983 -0.010644  \n",
-       "         ter                   0.591684  0.354459  \n",
-       "edittime bertscore                  NaN       NaN  \n",
-       "         bleu                       NaN       NaN  \n",
-       "         chrF                       NaN       NaN  \n",
-       "         editdist                   NaN       NaN  \n",
-       "         gptscore-noref-1-req       NaN       NaN  \n",
-       "         gptscore-ref-1-req         NaN       NaN  \n",
-       "         meteor                     NaN       NaN  \n",
-       "         rouge1                     NaN       NaN  \n",
-       "         rouge2                     NaN       NaN  \n",
-       "         rougeL                     NaN       NaN  \n",
-       "         ter                        NaN       NaN  "
-      ],
       "text/html": [
        "<div>\n",
        "<style scoped>\n",
@@ -1077,32 +1007,7 @@
        "  </tbody>\n",
        "</table>\n",
        "</div>"
-      ]
-     },
-     "execution_count": 47,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "execution_count": 47
-  },
-  {
-   "metadata": {
-    "ExecuteTime": {
-     "end_time": "2024-05-01T13:49:09.514129Z",
-     "start_time": "2024-05-01T13:49:09.295101Z"
-    }
-   },
-   "cell_type": "code",
-   "source": [
-    "from analysis_util import get_ref_only_correlations_for_groups\n",
-    "\n",
-    "get_ref_only_correlations_for_groups(df)"
-   ],
-   "id": "a3531f28722fa5bc",
-   "outputs": [
-    {
-     "data": {
+      ],
       "text/plain": [
        "                                    all              golden            \\\n",
        "                               spearman   pearson  spearman   pearson   \n",
@@ -1181,7 +1086,28 @@
        "         rouge2                     NaN       NaN  \n",
        "         rougeL                     NaN       NaN  \n",
        "         ter                        NaN       NaN  "
-      ],
+      ]
+     },
+     "execution_count": 47,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 50,
+   "id": "a3531f28722fa5bc",
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2024-05-01T13:49:09.514129Z",
+     "start_time": "2024-05-01T13:49:09.295101Z"
+    }
+   },
+   "outputs": [
+    {
+     "data": {
       "text/html": [
        "<div>\n",
        "<style scoped>\n",
@@ -1533,6 +1459,85 @@
        "  </tbody>\n",
        "</table>\n",
        "</div>"
+      ],
+      "text/plain": [
+       "                                    all              golden            \\\n",
+       "                               spearman   pearson  spearman   pearson   \n",
+       "relative independent                                                    \n",
+       "editdist bertscore            -0.184962 -0.129057 -0.316215 -0.254700   \n",
+       "         bleu                  0.260118  0.185995  0.269028  0.259690   \n",
+       "         chrF                 -0.199200 -0.129029 -0.343201 -0.300656   \n",
+       "         editdist              0.909934  0.910641  0.710772  0.662808   \n",
+       "         gptscore-noref-1-req  0.032048  0.055364  0.155510  0.048588   \n",
+       "         gptscore-ref-1-req    0.024550  0.035295 -0.009830 -0.062574   \n",
+       "         meteor                0.336016  0.371949  0.068034  0.173237   \n",
+       "         rouge1               -0.077574 -0.043738 -0.187349 -0.163230   \n",
+       "         rouge2                0.414256  0.340732  0.276139  0.332087   \n",
+       "         rougeL                0.006513 -0.008078 -0.041502 -0.034867   \n",
+       "         ter                   0.618095  0.385515  0.575614  0.501385   \n",
+       "edittime bertscore             0.140481  0.158807  0.140481  0.158807   \n",
+       "         bleu                  0.302380  0.326167  0.302380  0.326167   \n",
+       "         chrF                  0.079802  0.184202  0.079802  0.184202   \n",
+       "         editdist              0.252645  0.411131  0.252645  0.411131   \n",
+       "         gptscore-noref-1-req  0.206465  0.026235  0.206465  0.026235   \n",
+       "         gptscore-ref-1-req    0.130419 -0.055218  0.130419 -0.055218   \n",
+       "         meteor                0.253380  0.403564  0.253380  0.403564   \n",
+       "         rouge1                0.155926  0.136971  0.155926  0.136971   \n",
+       "         rouge2                0.218822  0.281944  0.218822  0.281944   \n",
+       "         rougeL                0.071344  0.091196  0.071344  0.091196   \n",
+       "         ter                   0.305601  0.062616  0.305601  0.062616   \n",
+       "\n",
+       "                                   +s2e                +e2s            \\\n",
+       "                               spearman   pearson  spearman   pearson   \n",
+       "relative independent                                                    \n",
+       "editdist bertscore            -0.308494 -0.113525 -0.181393 -0.165924   \n",
+       "         bleu                  0.512841  0.502827  0.109831  0.068138   \n",
+       "         chrF                 -0.238124 -0.064922 -0.233123 -0.201726   \n",
+       "         editdist              0.950494  0.935064  0.861930  0.878118   \n",
+       "         gptscore-noref-1-req  0.067857  0.047215 -0.029048 -0.013128   \n",
+       "         gptscore-ref-1-req   -0.015178 -0.036001  0.071345  0.087584   \n",
+       "         meteor                0.203616  0.425775  0.372598  0.360051   \n",
+       "         rouge1               -0.139874 -0.065543 -0.082093 -0.035603   \n",
+       "         rouge2                0.523559  0.537560  0.323911  0.282872   \n",
+       "         rougeL               -0.022288 -0.004664  0.012409  0.016372   \n",
+       "         ter                   0.774086  0.462554  0.529338  0.388592   \n",
+       "edittime bertscore                  NaN       NaN       NaN       NaN   \n",
+       "         bleu                       NaN       NaN       NaN       NaN   \n",
+       "         chrF                       NaN       NaN       NaN       NaN   \n",
+       "         editdist                   NaN       NaN       NaN       NaN   \n",
+       "         gptscore-noref-1-req       NaN       NaN       NaN       NaN   \n",
+       "         gptscore-ref-1-req         NaN       NaN       NaN       NaN   \n",
+       "         meteor                     NaN       NaN       NaN       NaN   \n",
+       "         rouge1                     NaN       NaN       NaN       NaN   \n",
+       "         rouge2                     NaN       NaN       NaN       NaN   \n",
+       "         rougeL                     NaN       NaN       NaN       NaN   \n",
+       "         ter                        NaN       NaN       NaN       NaN   \n",
+       "\n",
+       "                               +e2s+s2e            \n",
+       "                               spearman   pearson  \n",
+       "relative independent                               \n",
+       "editdist bertscore            -0.135421 -0.091748  \n",
+       "         bleu                  0.229712  0.145062  \n",
+       "         chrF                 -0.156914 -0.093376  \n",
+       "         editdist              0.939318  0.962305  \n",
+       "         gptscore-noref-1-req  0.012102  0.066882  \n",
+       "         gptscore-ref-1-req    0.013012  0.033618  \n",
+       "         meteor                0.392262  0.401802  \n",
+       "         rouge1               -0.054034 -0.030799  \n",
+       "         rouge2                0.433859  0.324538  \n",
+       "         rougeL                0.021983 -0.010644  \n",
+       "         ter                   0.591684  0.354459  \n",
+       "edittime bertscore                  NaN       NaN  \n",
+       "         bleu                       NaN       NaN  \n",
+       "         chrF                       NaN       NaN  \n",
+       "         editdist                   NaN       NaN  \n",
+       "         gptscore-noref-1-req       NaN       NaN  \n",
+       "         gptscore-ref-1-req         NaN       NaN  \n",
+       "         meteor                     NaN       NaN  \n",
+       "         rouge1                     NaN       NaN  \n",
+       "         rouge2                     NaN       NaN  \n",
+       "         rougeL                     NaN       NaN  \n",
+       "         ter                        NaN       NaN  "
       ]
      },
      "execution_count": 50,
@@ -1540,26 +1545,78 @@
      "output_type": "execute_result"
     }
    ],
-   "execution_count": 50
+   "source": [
+    "from analysis_util import get_correlations_for_groups\n",
+    "\n",
+    "get_correlations_for_groups(df, right_side=\"ind\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "d5dc33a4251baf9a",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "get_correlations_for_groups(df, right_side=\"aggr\")"
+   ]
+  },
+  {
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2024-05-01T15:25:18.226195Z",
+     "start_time": "2024-05-01T15:25:17.464762Z"
+    }
+   },
+   "cell_type": "code",
+   "source": [
+    "from matplotlib import pyplot as plt\n",
+    "\n",
+    "plt.scatter(x=df['edittime_related'], y=df['editdist_related'])"
+   ],
+   "id": "5df60ac60034b274",
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "<matplotlib.collections.PathCollection at 0x17c179da970>"
+      ]
+     },
+     "execution_count": 11,
+     "metadata": {},
+     "output_type": "execute_result"
+    },
+    {
+     "data": {
+      "text/plain": [
+       "<Figure size 640x480 with 1 Axes>"
+      ],
+      "image/png": "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"
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "execution_count": 11
   }
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "Python 3",
+   "display_name": "Python 3 (ipykernel)",
    "language": "python",
    "name": "python3"
   },
   "language_info": {
    "codemirror_mode": {
     "name": "ipython",
-    "version": 2
+    "version": 3
    },
    "file_extension": ".py",
    "mimetype": "text/x-python",
    "name": "python",
    "nbconvert_exporter": "python",
-   "pygments_lexer": "ipython2",
-   "version": "2.7.6"
+   "pygments_lexer": "ipython3",
+   "version": "3.9.5"
   }
  },
  "nbformat": 4,

analysis_util.py

@@ -1,6 +1,31 @@
+import functools
+import operator
+
 import pandas as pd
 
-from generation_steps.metrics_analysis import correlations_for_group
+
+def correlations_for_group(group):
+    REL_METRICS = [col.split("_")[0] for col in group.colmns if col.endswith("_related")]
+    IND_METRICS = [col.split("_")[0] for col in group.colmns if col.endswith("_independent")]
+    AGGR_METRICS = [col.split("_")[0] for col in group.colmns if col.endswith("_aggr")]
+
+    correlations = []
+    for rel_metric in REL_METRICS:
+        for ind_metric in IND_METRICS:
+            correlations.append({
+                f"rel_{rel_metric}_ind_{ind_metric}_pearson": group[f"{rel_metric}_related"].corr(
+                    group[f"{ind_metric}_independent"], method="pearson"),
+                f"rel_{rel_metric}_ind_{ind_metric}_spearman": group[f"{rel_metric}_related"].corr(
+                    group[f"{ind_metric}_independent"], method="spearman"),
+            })
+        for aggr_metric in AGGR_METRICS:
+            correlations.append({
+                f"rel_{rel_metric}_aggr_{aggr_metric}_pearson": group[f"{rel_metric}_related"].corr(
+                    group[f"{aggr_metric}_aggr"], method="pearson"),
+                f"rel_{rel_metric}_aggr_{aggr_metric}_spearman": group[f"{rel_metric}_related"].corr(
+                    group[f"{aggr_metric}_aggr"], method="spearman"),
+            })
+    return pd.Series(functools.reduce(operator.ior, correlations, {}))
 
 
 def split_metrics_string(s):
@@ -8,10 +33,10 @@ def split_metrics_string(s):
     return tokens[1], tokens[3]
 
 
-def get_ref_only_correlations_df(df):
+def get_correlations_df(df, right_side):
     correlations_raw = correlations_for_group(df)
 
-    idx = list(set("_".join(col.split("_")[:-1]) for col in correlations_raw.index))
+    idx = list(set("_".join(col.split("_")[:-1]) for col in correlations_raw.index if right_side in col))
 
     data = []
     for metrics in idx:
@@ -29,8 +54,8 @@ def get_ref_only_correlations_df(df):
     return result
 
 
-def get_ref_only_correlations_for_groups(df):
-    noref_correlations = {"all": get_ref_only_correlations_df(df)}
+def get_correlations_for_groups(df, right_side):
+    noref_correlations = {"all": get_correlations_df(df, right_side=right_side)}
 
     for e2s in (False, True):
         for s2e in (False, True):
@@ -43,7 +68,7 @@ def get_ref_only_correlations_for_groups(df):
                 suffix = "golden"
 
             subdf = df[(df["end_to_start"] == e2s) & (df["start_to_end"] == s2e)]
-            subdf_noref_corr = get_ref_only_correlations_df(subdf)
+            subdf_noref_corr = get_correlations_for_groups(subdf, right_side=right_side)
             noref_correlations[suffix] = subdf_noref_corr
 
     noref_correlations = pd.concat(noref_correlations, axis=1)

change_visualizer.py

@@ -108,7 +108,10 @@ if __name__ == '__main__':
                     layout_for_statistics("synthetic")
 
             gr.Markdown(f"### Reference-only correlations")
-            gr.Markdown(value=analysis_util.get_ref_only_correlations_for_groups(df_synthetic).to_markdown())
+            gr.Markdown(value=analysis_util.get_correlations_for_groups(df_synthetic, right_side="ind").to_markdown())
+
+            gr.Markdown(f"### Aggregated correlations")
+            gr.Markdown(value=analysis_util.get_correlations_for_groups(df_synthetic, right_side="aggr").to_markdown())
 
         application.load(update_dataset_view_manual, inputs=slider_manual,
                          outputs=view_manual)

generation_steps/metrics_analysis.py

@@ -7,6 +7,7 @@ import pandas as pd
 from tqdm import tqdm
 
 import config
+from analysis_util import correlations_for_group
 from api_wrappers import hf_data_loader
 from custom_metrics import gpt_eval
 
@@ -110,6 +111,10 @@ IND_METRICS = {
     "ter": ter_fn,
 }
 
+AGGR_METRICS = IND_METRICS.copy()
+del AGGR_METRICS["gptscore-ref-1-req"]
+del AGGR_METRICS["gptscore-noref-1-req"]
+
 REL_METRICS = {
     "editdist": edit_distance_fn,
     "edittime": edit_time_fn,
@@ -128,6 +133,22 @@ def compute_metrics(df):
     def apply_metric_fn_to_row(row, fn, col_pred, col_ref):
         return fn(row[col_pred], row[col_ref], edittime=row['edit_time'], diff=str(row['mods']))
 
+    for metric in AGGR_METRICS:
+        print(f"Computing {metric} for the aggregated independent pairs")
+        values = []
+        for i, row in tqdm(df.iterrows(), total=len(df)):
+            others = df[(df["hash"] == row["hash"]) & (df["repo"] == row["repo"]) & (
+                    df["commit_msg_start"] != row["commit_msg_start"])]['commit_msg_end'].to_list()
+            others.append(row["reference"])
+            others = list(set(others))
+            metric_fn = AGGR_METRICS[metric]
+            values.append(
+                metric_fn(
+                    row['commit_msg_start'], None, refs=others, edittime=row['edit_time'], diff=str(row['mods'])
+                )
+            )
+        df[f"{metric}_aggr"] = values
+
     for metric in REL_METRICS:
         print(f"Computing {metric} for the related pairs")
         metric_fn = REL_METRICS[metric]
@@ -158,24 +179,15 @@ def compute_metrics(df):
             df[f"rel_{rel_metric}_ind_{ind_metric}_spearman"] = (
                 df[f"{rel_metric}_related"].corr(df[f"{ind_metric}_independent"], method="spearman"))
 
-    return df
+        for aggr_metric in AGGR_METRICS:
+            df[f"rel_{rel_metric}_aggr_{aggr_metric}_pearson"] = (
+                df[f"{rel_metric}_related"].corr(df[f"{aggr_metric}_aggr"], method="pearson"))
 
+            df[f"rel_{rel_metric}_ind_{aggr_metric}_spearman"] = (
+                df[f"{rel_metric}_related"].corr(df[f"{aggr_metric}_aggr"], method="spearman"))
+
+    return df
 
-def correlations_for_group(group):
-    correlations = []
-    for rel_metric in REL_METRICS:
-        # correlations.append({
-        #     f"{metric}_pearson": group[f"{metric}_related"].corr(group[f"{metric}_independent"], method="pearson"),
-        #     f"{metric}_spearman": group[f"{metric}_related"].corr(group[f"{metric}_independent"], method="spearman")
-        # })
-        for ind_metric in IND_METRICS:
-            correlations.append({
-                f"rel_{rel_metric}_ind_{ind_metric}_pearson": group[f"{rel_metric}_related"].corr(
-                    group[f"{ind_metric}_independent"], method="pearson"),
-                f"rel_{rel_metric}_ind_{ind_metric}_spearman": group[f"{rel_metric}_related"].corr(
-                    group[f"{ind_metric}_independent"], method="spearman"),
-            })
-    return pd.Series(functools.reduce(operator.ior, correlations, {}))
 
 
 def compute_correlations(df: pd.DataFrame):

requirements.txt

@@ -63,7 +63,6 @@ jsonpointer==2.4
 jsonschema==4.21.1
 jsonschema-specifications==2023.12.1
 kiwisolver==1.4.5
-Levenshtein==0.25.1
 lxml==5.2.1
 markdown-it-py==3.0.0
 MarkupSafe==2.1.5