Hive 查询：如何使用分组依据排名？

Question

I have a table like below

year                            int                                                                                                                   
month                           int                                                                                                                   
symbol                          string                                                                                                                
company_name                    string                                                                                                                
sector                          string                                                                                                                
sub_industry                    string                                                                                                                
state                           string                                                                                                                
avg_open                        double                                                                                                                
avg_close                       double                                                                                                                
avg_low                         double                                                                                                                
avg_high                        double                                                                                                                
avg_volume                      double         

The field starting with avg_ refers to the average value in a month for a year. I need to find for each sector the year in which average of avg_close is the lowest.

I tried to do something like below

SELECT sector, year FROM
  (
    SELECT sector, year, RANK() OVER (ORDER BY s2.yearly_avg_close) AS RANK FROM
      ( SELECT year,sector, AVG(avg_close) AS yearly_avg_close FROM stock_summary GROUP BY sector, year) s2
  ) s1 
WHERE
  s1.RANK = 1;

But this is printing just one sector and year like below

Telecommunications Services     2010

I am new to hive and playing around with some toy schemas. Can someone let me know what should be the correct way of solving this?

Hive Version - 1.1.0

Answer 1

Include sector into the partition by in the rank() function:

SELECT sector, year, RANK() OVER (partition by sector ORDER BY s2.yearly_avg_close) AS RANK

Add year as well if you need rank per each sector and year

Read also this explanation how rank works: https://stackoverflow.com/a/55909947/2700344