如何更优雅地按列分组频次过滤Kdb+数据表？

Great question! Your current approach gets the job done, but kdb+ has some idiomatic shortcuts that make this task cleaner and more concise. Let's look at a few better options:

• Single-pass filter with fby (my top pick)
  The fby (filter-by) operator is made for this kind of task—it lets you calculate group-level stats inline without precomputing a separate grouping dictionary. This one-liner does everything in a single table scan:

  select from tmp where 1 < count i fby id
  

  How it works: count i fby id computes the total number of rows for each id group and attaches that count to every row in the group. We then just filter for rows where that count is greater than 1. No extra variables needed!

• Group-and-raze method
  If you prefer working directly with grouped data, this approach leverages group and raze to extract only the groups with multiple rows:

  raze where 1 < count each group tmp[`id]
  

  Breakdown: group tmp[id]creates a dictionary where keys areidvalues and values are the corresponding rows.count eachgets the size of each group,where 1 < ...keeps only groups with more than one row, andraze` flattens those groups back into a single table.

• Add a count column (if you need to keep frequency data)
  If you want to retain the group count alongside your filtered rows, use update with fby to add the count first, then filter:

  update groupCount:count i fby id from tmp where groupCount > 1
  

  This gives you the filtered rows and shows how many times each id appeared, which can be handy for debugging or further analysis.

All these methods skip the extra step of storing a separate ce variable, making your code more compact and aligned with kdb+'s functional style.

内容的提问来源于stack exchange，提问作者tenticon