• Status: Solved
• Priority: Medium
• Security: Public
• Views: 432

# Filter by Row in Excel (Part 2 - Efficiency!)

I previously asked a query on Filtering by a value in a Row, rather than a column:

http://www.experts-exchange.com/Software/Office_Productivity/Office_Suites/MS_Office/Excel/Q_28363142.html#a39856514

"I would like a way, if possibly, to filter through this by ROW, rather than Column.

For example, in each group we have a Row labelled Responsibility Application. I want to Filter by Responsibility Application ."

This was answered perfectly by user: byundt

Attached is a working example:  ExpertsExchangeExample2Q28363142.xlsm

This solution works perfectly for a dataset of about 5000 rows or less. It takes 40-50 seconds to calculate, but works and is usable.

However, I have been given the same task on a file with 25,000 rows. This solution of filtering by Row causes Excel and the laptops in question to almost break down, so unfortunately can be used.

Can anyone, byundt included, come up with a way to run this or a solution that will work on very large data sets?

Thank you,

Alex
0
EllenExcel
4 Solutions

Author Commented:
I'm attaching a blank example (with just data, not the previous solution):

ExpertsExchangeExample2.xlsm.xlsx
0

Commented:
Smart solution.

Computations for entire columns take a long time.
If the search is narrowed down to just done when a new section starts,
(Col A=Responsibility), and only looking at a smaller range.

For column D, the numbering.
I have changed to numbers, to use the Max function, only performed when a new block start.
For D6, copied down.
``````=IF(A6="","",IF(A6="Responsibility",MAX(D\$5:D5)+1,D5))
``````

For Column E, Result True/False, only looking 100 rows down
For E6, copied down
``````=IF(D6="","",IF(A6="Responsibility",(COUNTIFS(D6:D105,D6,A6:A105,A\$2,B6:B105,B\$2)+COUNTIFS(D6:D105,D6,A6:A105,A\$2,C6:C105,B\$2))>0,E5))
``````

With 4.500 rows it just takes a second.
See file
Filter-by-row.xlsm
0

Commented:
Another approach, relying wholly on VBA:

Q-28366250.xlsm

Use A2:B2 to change what field you filter, and what value you seek for that field.

Note the following:
Code looks at key values in Columns B:???.  To control how many columns are being used, change the NumColumns constant
If the value being sought is found anywhere in those columns, the item is kept (i.e., not hidden)
0

Commented:
Alex,
I had thought of restricting the range in the MAX and COUNTIFS functions in your original question, but chose not to in the interest of simplicity. To take advantage of these formula improvements, you would need to specify both the maximum "gap" between records and the maximum number of rows possible in a "record".

hgholt has already suggested a suitable improvement for the COUNTIFS formula in column E. He assumed no more than 100 rows in a record.

Here is an improved MAX formula in cell D6 (should be copied down). It assumes there are no more than four blank rows between records.
=IF(A6="","",IF(A6="Responsibility",MAX(D1:D5)+1,D5))

With 26,000 rows of data, the macro found a single record located near the end of the list in a couple of seconds on my six  year old Macintosh running Windows in emulation. No change in macro code was necessary.

If this were my workbook, I would be thinking about putting each record on a single row. The more you analyze the data, the more firmly I feel such a change would be needed. Putting each record on a single row would mean that each of the different fields would be in its own column. A macro to remap the data wouldn't be hard to write.

0

Commented:
See attached.

Type the FieldName by which you want to filter in B1 i.e. "Responsibility Name"
column D will be populated.

Use filter in D3 to filter by value of the FieldName you have selected.

The formula for populating D, in D4 is
``````=IF(A4="Responsibility", VLOOKUP(\$B\$1, A4:B999,2,FALSE), D3)
``````

Here is the explanation for the same:
A4="Responsibility" => Detect first row of the block of of rows constituting one record
VLOOKUP(\$B\$1, A4:B999,2,FALSE) => Value of selected FieldName in this record. 999 here is arbitrary, should be greater or equal to max no of rows in a record.
D3 => Cell above

Logic: If it is start of the block, then set cell to value of selected key in the record, for subsequent rows, simply copy the value set above.

Note: If selected field is not present in a record but present in subsequent record that is covered in next 999 rows, then result will be erroneous.
There is more complex formula in column E which solves this issue but has problem with last record. This can be solved by ensuring an empty record at the end, i.e. insert "Responsibility" in in A142.
Filter-By-Row-Solution.xls
0

Author Commented:
All of your solutions work, so I've spread the points across the lot.

Thanks all for the comments - Brad you are right it would be much simpler to transpose everything up to a column view looking at it.

I wanted this solution because who this is for preferred to keep the sheet in the same format as the 3rd party product produces, so I was restricted to working with that and fell short with the know-how.

Thank you all,

Alex
0

Commented: