From c1c2761c7985b971cc60e36064928459249f50e6 Mon Sep 17 00:00:00 2001
From: Franck Dary <franck.dary@lis-lab.fr>
Date: Thu, 18 Feb 2021 10:16:11 +0100
Subject: [PATCH] Added option mean to printResults
---
UD_any/print_results.py | 54 ++++++++++++++++++++++++++++++++++-------
1 file changed, 45 insertions(+), 9 deletions(-)
diff --git a/UD_any/print_results.py b/UD_any/print_results.py
index e1eeb68..21a4d94 100755
--- a/UD_any/print_results.py
+++ b/UD_any/print_results.py
@@ -1,12 +1,25 @@
#! /usr/bin/env python3
+import argparse
import glob
import sys
import math
+import copy
################################################################################
if __name__ == "__main__" :
+ parser = argparse.ArgumentParser()
+ parser.add_argument("dir", type=str, default="",
+ help="Directory containing the .stdout (scores) files.")
+ parser.add_argument("--score", "-s", type=str, default="F1",
+ help="Name of the score to report (F1,L1,L2,R2).")
+ parser.add_argument("--metrics", type=str, default="Sentences,LAS,UAS,Lemmas,UFeats,UPOS,Words,Tokens",
+ help="Comma separated list of metrics.")
+ parser.add_argument("--mean", "-m", default=False, action="store_true",
+ help="compute the mean of metrics.")
+ args = parser.parse_args()
+
scoreTypes = {
"F1" : ["F1.score","%",-1,"%.2f"],
"R2" : ["R²","",-3,"%.4f"],
@@ -14,15 +27,17 @@ if __name__ == "__main__" :
"L2" : ["L2","",-2,"%.2f"],
}
- scoreType = scoreTypes["F1"] if len(sys.argv) < 3 else scoreTypes[sys.argv[2].upper()]
+ scoreType = scoreTypes[args.score.upper()]
- metrics = ["Sentences","LAS","UAS","Lemmas","UFeats","UPOS","Words","Tokens"]
+ metrics = args.metrics.split(',')
+ usedMetrics = []
+ mean = args.mean
output = []
outputByModelScore = dict()
- filenamesErr = glob.iglob((sys.argv[1]+"/"if len(sys.argv) > 1 else "") + '*stderr')
- filenamesOut = glob.iglob((sys.argv[1]+"/"if len(sys.argv) > 1 else "") + '*stdout')
+ filenamesErr = glob.iglob((args.dir+"/"if len(args.dir) > 1 else "") + '*stderr')
+ filenamesOut = glob.iglob((args.dir+"/"if len(args.dir) > 1 else "") + '*stdout')
for pathToFile in filenamesErr :
for line in open(pathToFile, "r") :
@@ -46,10 +61,31 @@ if __name__ == "__main__" :
if model not in outputByModelScore[corpus] :
outputByModelScore[corpus][model] = dict()
- if splited[0] not in outputByModelScore[corpus][model] :
- outputByModelScore[corpus][model][splited[0]] = []
-
- outputByModelScore[corpus][model][splited[0]].append([corpus, splited[0], splited[scoreType[2]], model])
+ if metric not in outputByModelScore[corpus][model] :
+ outputByModelScore[corpus][model][metric] = []
+ if metric not in usedMetrics :
+ usedMetrics.append(metric)
+
+ outputByModelScore[corpus][model][metric].append([corpus, metric, splited[scoreType[2]], model])
+
+ for metric in metrics :
+ if metric not in usedMetrics :
+ print("WARNING : Unused metric '%s'"%metric, file=sys.stderr)
+
+ if mean :
+ metricName = ",".join(metrics)
+ for corpus in outputByModelScore :
+ for model in outputByModelScore[corpus] :
+ nbRedo = len((list(outputByModelScore[corpus][model].values())[0]))
+ newMetrics = copy.deepcopy(list(outputByModelScore[corpus][model].values())[0])
+ for elem in newMetrics :
+ elem[2] = 0
+ elem[1] = metricName
+ for redo in range(nbRedo) :
+ for metric in outputByModelScore[corpus][model] :
+ newMetrics[redo][2] += float(outputByModelScore[corpus][model][metric][redo][2])
+ newMetrics[redo][2] /= len(metrics)
+ outputByModelScore[corpus][model] = {metricName : newMetrics}
for corpus in outputByModelScore :
for model in outputByModelScore[corpus] :
@@ -73,7 +109,7 @@ if __name__ == "__main__" :
score = score.replace('-','')
output.append(outputByModelScore[corpus][model][metric][0])
output[-1][2] = score
- output[-1] = [output[-1][0]] + [metrics.index(output[-1][1])] + [output[-1][1]] + [baseScore] + output[-1][2:]
+ output[-1] = [output[-1][0]] + ([0] if mean else [metrics.index(output[-1][1])]) + [output[-1][1]] + [baseScore] + output[-1][2:]
if len(output) == 0 :
print("ERROR : Output length is 0", file=sys.stderr)
--
GitLab