The combination of kvgen and flatten would be the best approach. kvgen
returns a repeated map, flatten will convert the repeated map into separate
single maps. You can then aggregate or filter on the key or value.
One problem with this, however, is that in this particular example, it
looks like yo
I did use kvgen, but not sure how. The price value is a long , how can I
calculate an average for this , when the column names are variable?
0: jdbc:drill:> select convert_from(row_key, 'UTF8') as tid, kvgen(t.price)
as price from dfs.`/tables/trades_flat` t limit 5;
+++
|
Recently got this question on how to extract the day of the week from date.
Here is one method.
select datecolumn, to_char(datecolumn, 'E') from base_table limit 10;
+++
|date| EXPR$1 |
+++
| 2015-02-01 | Sun|
| 2015-02-01 | Sun
Would kvgen work on t.price?
On Thu, Feb 19, 2015 at 12:59 PM, Carol McDonald
wrote:
> What is the best way to query an hbase table that has dynamic column names
> ? For example this table is similar to the opentsdb table, the rowkey is a
> stocksymbol followed by the date and hour , the Price
