Shuffling is important

1 files changed, 11 insertions, 1 deletions

diff --git a/extract.sh b/extract.sh
index 77c0ddc..e0cb691 100755
--- a/extract.sh
+++ b/extract.sh
@@ -36,7 +36,17 @@ process() {
 		shuf -n $num_games "data/good-playtak-$size" > "data/smalltak-$size"
 		echo -en "\tGenerating training data... "
 		./pptdb "$size" "data/smalltak-$size" generate
-		echo -e "\n\tDone! Wrote $(wc -l data/training-$size.csv | cut -d\  -f1) samples to data/training-$size.csv."
+		echo -e "\tDone, $(wc -l data/training-$size.csv | cut -d\  -f1) samples."
+		echo -en "\tShuffling... "
+		head -n 1 "data/training-$size.csv" > "data/shuf-$size.csv"
+		tail -n+2 "data/training-$size.csv" | shuf >> "data/shuf-$size.csv"
+		mv "data/shuf-$size.csv" "data/training-$size.csv"
+		echo -en "Done.\n\tCompressing data... "
+		if [ -f "data/training-$size.csv.gz" ]; then
+				rm "data/training-$size.csv.gz"
+		fi
+		gzip "data/training-$size.csv"
+		echo -e "Done!\n\tCompressed version available in data/training-$size.csv.gz"
 }