重塑Postgres桌子，长到宽

Question

Considering the following time series data in Postgres 9.5.1, I am trying to reshape to an array(int) |time_to_failure(double) as new table

# SELECT acoustic_data, time_to_failure FROM lanl_train WHERE 
time_to_failure = '0.000795';
 acoustic_data | time_to_failure 
---------------+-----------------
             2 |        0.000795
             5 |        0.000795
             6 |        0.000795
             1 |        0.000795
             2 |        0.000795
             5 |        0.000795
             8 |        0.000795
             5 |        0.000795
             4 |        0.000795
             4 |        0.000795
             7 |        0.000795
             2 |        0.000795
             2 |        0.000795
             0 |        0.000795
             0 |        0.000795
             2 |        0.000795
             4 |        0.000795
             5 |        0.000795
             4 |        0.000795
(19 rows)

# SELECT ARRAY(SELECT acoustic_data FROM lanl_train WHERE 
time_to_failure = '0.000795');
                  array                  
-----------------------------------------
 {2,5,6,1,2,5,8,5,4,4,7,2,2,0,0,2,4,5,4}
(1 row)

So that a row in new table would be

acoustic_data(array) | time_to_failure(double)
----------------------------------------------
{2,5,6,1,2,5,8,5,4,4,7,2,2,0,0,2,4,5,4} | 0.000795

I have some of the parts but am stuck on SELECT to achieve this result. Any help greatly appreciated.

Answer 1

You want array_agg ( docs ) with a group by:

select
    array_agg(lanl_train.acoustic_data) as acoustic_data,
    lanl_train.time_to_failure
from
    lanl_train
where
    lanl_train.time_to_failure = '0.000795'
group by
    lanl_train.time_to_failure