There is no need to use cat to read the file. AWK is perfectly capable to read it.

A core c[$3]++ statement should get the count of lines of each type.
Then, at the end, just print (as tab separated values) all the counts:

#!/bin/bash

awk -F '\t' '   {c[$3]++}
                 END{
                     for (i in c) printf("%s\t%s\n",i,c[i])
                 }' dataset.csv 

Appended

Given the comment from the OP that:

I get some issues for colums that have quotes like that doesn\'t mean that you\'re not worth remembering think of the people who need to know they need to know so you need ​to show.... In this case the parsing on \t will fail.

I got to review the answer. I created this file:

$ cat dataset.csv 
1233    that doesn\'t mean that you\'re not worth remembering think of the people who need to know they need to know so you need to show...    CLASS_0
1234    here    CLASS_A
1235    goes the values CLASS_B
1236    "that need counting"    CLASS_B
1237    "\like \this"   CLASS_B
1238    \or \this       CLASS_C
1239    including spaces        CLASS_B
1240    but not tabs    CLASS_A
1241    which could not work    CLASS_B
1242    finally CLASS_C
1243    this is CLASS_A
1244    over    CLASS_B
1245    988     CLASS_C

That file, when used with the script, gives the correct result:

$ ./script
CLASS_A 3
CLASS_B 6
CLASS_C 3
CLASS_0 1

Which is the correct result.

Of course, the file

1. has the correct amount of tabs for 3 fields, and
2. variables are correctly quoted when expanded and are not in upper case.

To test that a file does comply with the first requirement, you may use this script:

#!/bin/bash

filetoread="$2"

<"$filetoread" tr -dc '\t\n' |
    awk '(length!=2){printf("Error in line: %s, has %s tabs\n",NR,length)}'

awk -F '\t' '(NF!=3){printf("Error in line: %s, has %s fields\n",NR,NF)}' "$filetoread"

Which checks that there are exactly two tabs per line, and
That the number of fields (as seen by awk) are actually three.

Adding a couple of test lines:

… …
1239    including spaces        CLASS_B
1       but not     tabs    CLASS_A
2       but not \ttabs  CLASS_A
1240    but not tabs    CLASS_A
… …

And running the script above:

$ ./script 3 dataset.csv
Error in line: 8, has 4 tabs
Error in line: 8, has 5 fields

detects the line ID 1 that has four tabs (two added) and doesn't get fooled by line ID 2 with a \t.

As for the quoting and use of variables, that is something you should improve all by yourself.

GNU datamash

$ datamash -t, count 3 sum 3 < file
3,604720

Some testing

$ time gawk -F',' '{ sum += $3 } END{ print sum, NR }' longfile
604720000000 3000000

real    0m2.851s
user    0m2.784s
sys     0m0.068s

$ time mawk -F',' '{ sum += $3 } END{ print sum, NR }' longfile
6.0472e+11 3000000

real    0m0.967s
user    0m0.920s
sys     0m0.048s

$ time perl -F, -nle '$sum += $F[2] }{ print "$.,$sum"' longfile
3000000,604720000000

real    0m3.394s
user    0m3.364s
sys     0m0.036s

$ time { cut -d, -f3 <longfile |paste -s -d+ - |bc ; }
604720000000

real    0m1.679s
user    0m1.416s
sys     0m0.248s

$ time datamash -t, count 3 sum 3 < longfile
3000000,604720000000

real    0m0.815s
user    0m0.716s
sys     0m0.036s

So mawk and datamash appear to be the pick of the bunch.