Performance Tuning :: Cannot Collect Top Frequency Histogram On Column?

I'm trying to collect histograms for column COL_C of table TAB_A(150K records), So an index "BAD_IDX" will *not* be used in a query when the value is not selective.

This is my query:

SELECT COL_A, COL_B , COL_C , COL_D , COL_E , COL_F
FROM TAB_A
WHERE COL_A = 050
AND COL_B = 13012345
AND COL_C = 0
AND COL_D = 0
AND COL_D >= '07/23/2013 00:00:00'
ORDER BY COL_E ASC;

Now, I have index "BAD_IDX" on columns (COL_C, COL_E ).and the distribution of values looks like this:

select COL_C, count(*)
FROM TAB_A --very not selective for 0, selective for the rest, also no histogram
group by rollup(COL_C)
order by 2 desc;

and the result is 20k row long (20k distincts), So I'll post just the top part of it:

COL_CCOUNT(*)
-------------
158382
086356
695583772
623044169
659558964
680045564
689664263

[code]...

Now, the problem with the query was that "COL_B = 13012345" was the most selective predicate, And an index for it did not exist, so the index "BAD_IDX" is used, and is scanning 86k records (all the "0" value records for column COL_C)!

So, I created an index

Create index GOOD_IDX on TAB_A(COL_B) compute statistics;

However, that BAD_IDX index is still being used! I've thought that maybe it's because the lack of histograms for the column COL_C.I've also understood from documentation I've read that the suitable histogram type is TOP FREQUENCY, Because although I have 20k distincts here for the column COL_C, what does the difference is the 86k records of value 0.

I try to create histograms for the column:

BEGIN
DBMS_STATS.GATHER_TABLE_STATS (
ownname => 'ANDREY'
, tabname => 'TAB_A'
, method_opt => 'FOR COLUMNS COL_C'
);
END;
/

But then when I check - I see that

1. My query is still using the wrong index

2. My histogram is height-balanced and not FREQUENCY

select histogram from dba_tab_columns where table_name='TAB_A' and column_name='COL_C' and owner=user;

The result is:

HISTOGRAM
----------
HEIGHT BALANCED

What do I need to do in order to

1. Make my query use the correct index

2. Create a FREQUENCY histogram