script can now create 8_1_1 datasets

data.txt

@@ -0,0 +1,2 @@
+Archive:  wine.zip
+  inflating: winequality-red.csv     

script.sh

@@ -1,4 +1,56 @@
 #!/bin/sh
-kaggle datasets download -d uciml/red-wine-quality-cortez-et-al-2009
+#kaggle datasets download -d uciml/red-wine-quality-cortez-et-al-2009
-unzip -o red-wine-quality-cortez-et-al-2009.zip > data.txt
+#unzip -o red-wine-quality-cortez-et-al-2009.zip > data.txt
-head data.txt
+
+
+FILE=winequality-red.csv
+
+TRAIN_SET_PERCENT=80
+TEST_SET_PERCENT=10
+VAL_SET_PERCENT=$(( 100 - $TRAIN_SET_PERCENT - $TEST_SET_PERCENT ))
+
+TRAIN_DATA=TRAIN_$FILE
+TEST_DATA=TEST_$FILE
+VAL_DATA=VAL_$FILE
+
+> $TRAIN_DATA
+> $TEST_DATA
+> $VAL_DATA
+
+for state in `cut -d',' -f1 $FILE | sort | uniq`
+do
+    NUM_STATE_DATA=`grep "$state" $FILE | wc -l`
+    echo "$state: $NUM_STATE_DATA"
+
+    TRAIN_NUM_DATA=$(( $NUM_STATE_DATA * $TRAIN_SET_PERCENT / 100 ))
+    TEST_NUM_DATA=$(( $NUM_STATE_DATA * $TEST_SET_PERCENT / 100 ))
+    VAL_NUM_DATA=$(( $NUM_STATE_DATA - $TRAIN_NUM_DATA - $TEST_NUM_DATA ))
+
+    
+    STATE_DATA=`grep $state $FILE`
+    
+
+    # Train set
+    per=$TRAIN_SET_PERCENT
+    num=$TRAIN_NUM_DATA; from=1; to=$(($from + $num - 1));
+    echo Train set: $per% $num from=$from to=$to
+    echo "$STATE_DATA" | head -$to >> $TRAIN_DATA
+
+    # Test set
+    per=$TEST_SET_PERCENT
+    num=$TEST_NUM_DATA; from=$(($to + 1)); to=$(($from + $num - 1));
+    echo Test set: $per% $num from=$from to=$to
+    echo "$STATE_DATA" | head -$to | tail -$num >> $TEST_DATA
+
+    # Validate set
+    per=$VAL_SET_PERCENT
+    num=$VAL_NUM_DATA; from=$(($to + 1)); to=$NUM_STATE_DATA;
+    echo Validate set: $per% $num from=$from to=$to
+    echo "$STATE_DATA" | tail -$num >> $VAL_DATA
+
+    echo
+done
+
+
+
+