# Count all members in an array field with counter aggregator for whole table **URL:** \ **Category:** Hail Query & hailctl\ **Created:** [August 28, 2021, 9:19pm UTC](https://discuss.hail.is/t/count-all-members-in-an-array-field-with-counter-aggregator-for-whole-table/2210 "2021-08-28T21:19:19Z")\ **Posts on this page:** 4\ **Page:** 1
**Author:** ![thondeboer](https://yyz2.discourse-cdn.com/flex036/user_avatar/discuss.hail.is/thondeboer/32/352_2.png) [@thondeboer](https://discuss.hail.is/u/thondeboer)\ **Post date:** [August 28, 2021, 9:19pm UTC](https://discuss.hail.is/t/count-all-members-in-an-array-field-with-counter-aggregator-for-whole-table/2210/1 "2021-08-28T21:19:19Z")
I have an entry field in a matrix table with two string values and I want to count them with the hl.agg.acounter but cannot figure out how to do this for each entry in the array. ```auto ht = hl.utils.range_table(10) ht = ht.annotate(arr = hl.range(0, 2).map(lambda _: hl.if_else(hl.rand_bool(0.5),'REF','PTV'))) ht.show() ``` ```auto idx. arr int32 array 0 ["REF","PTV"] 1 ["PTV","REF"] 2 ["REF","REF"] 3 ["PTV","PTV"] 4 ["PTV","PTV"] 5 ["REF","REF"] 6 ["PTV","REF"] 7 ["REF","REF"] 8 ["REF","REF"] 9 ["PTV","PTV"] ``` I would like to be able to count each occurance of ‘REF’ and ‘PTV’ with hl.agg.counter or something so I could get this: ```auto {'PTV': 9, 'REF': 11} ``` I get close with something like this: ```auto ht.aggregate(hl.agg.array_agg(lambda x: hl.agg.counter(x),ht.arr)) [{'PTV': 5, 'REF': 5}, {'PTV': 4, 'REF': 6}] ``` But then I need to run through that list and add each keys value etc… Or this, but then I need to know the actual field values, which I don’t ```auto ht.aggregate( hl.dict([ ('REF', hl.agg.sum(ht.arr.filter(lambda x: x == 'REF').size())), ('PTV', hl.agg.sum(ht.arr.filter(lambda x: x == 'PTV').size())) ]) ) {'PTV': 9, 'REF': 11} ``` Is there an easier way to do this? ---
**Author:** ![tpoterba](https://yyz2.discourse-cdn.com/flex036/user_avatar/discuss.hail.is/tpoterba/32/61_2.png) [@tpoterba](https://discuss.hail.is/u/tpoterba)\ **Post date:** [August 28, 2021, 11:56pm UTC](https://discuss.hail.is/t/count-all-members-in-an-array-field-with-counter-aggregator-for-whole-table/2210/2 "2021-08-28T23:56:00Z")
This is exactly what [`hl.agg.explode`](https://hail.is/docs/0.2/aggregators.html#hail.expr.aggregators.explode) is for! ```auto ht.aggregate(hl.agg.explode(lambda element: hl.agg.counter(element), ht.arr)) ``` ---
**Author:** ![thondeboer](https://yyz2.discourse-cdn.com/flex036/user_avatar/discuss.hail.is/thondeboer/32/352_2.png) [@thondeboer](https://discuss.hail.is/u/thondeboer)\ **Post date:** [August 29, 2021, 12:31am UTC](https://discuss.hail.is/t/count-all-members-in-an-array-field-with-counter-aggregator-for-whole-table/2210/3 "2021-08-29T00:31:57Z")
I had experimented with Explode but did not get anywhere. This works great! ---
**Author:** ![tpoterba](https://yyz2.discourse-cdn.com/flex036/user_avatar/discuss.hail.is/tpoterba/32/61_2.png) [@tpoterba](https://discuss.hail.is/u/tpoterba)\ **Post date:** [August 29, 2021, 12:35am UTC](https://discuss.hail.is/t/count-all-members-in-an-array-field-with-counter-aggregator-for-whole-table/2210/4 "2021-08-29T00:35:28Z")
The “control flow” aggregators are `filter` (aggregate records conditionally, according to some predicate), `explode` (apply an aggregation to each element of a collection), `group_by` (apply an aggregation to each record, grouped according to some other value, and `array_agg` (aggregate elements of a fixed-size array by position, returning an array of results for each index). These can be nested arbitrarily: ```auto hl.agg.filter(PREDICATE, hl.agg.group_by(KEY, hl.agg.explode(lambda x: INNER_AGG, ARRAY)) ``` is totally valid.