Need a better optimized code?

Question

Need a much Optimized code.Well I Got a Project and I have Succefully made it work with the vba (Mostly helped by the stackoverflow programmers Thanks for that) But Today I got a Feedback. Its deleting 2 more unique entries in the record But I dont know why its deleting Them.

The Algorithm I have applied

I have Used the COUNTIF function Which I found on google

    ="countif(A$1:A2,A3)=0" A3 is the active cell, Checks A2,A1 for dupes

It Throws False if there is a duplicate in The A column and True If it is a unique.What I have understood about Countif is that It checks all the above columns values from that cell I mean let us take A4. SO it checks A2,A1,A3 for the duplicate. Similarly A10 checks for A1 to A9 and throws either TRue or False.Well It was working But I dont know what went wrong The code is not working for some entries.Its even showing False for the Unique entries sometimes.

And its taking more time to applye these formula as I have more amount of data. Im trying to make it cleaner and more Optimizing Way.People told me its not a c or some other Language to make it optimize but Im need of code that makes my code more optimized

I need code for these condtions can anyone help me as my countif failed.Im little helpless in doing so.

1)I have a column and I should check for duplicates in that column and delete that row if it is a duplicate

2) I have 35000 Old entries in the column and I have new entries 2000 everyweek these are appended. I need to check these 2000 entries from the total 37000 ( as we appened we get 35000+2000) and these delete operation need to be performed only on the newly appended 2000 entries but it should check the duplicates for entire column

Let me explain you clearly I have 2000 entries newly added,so Only these entries are to be checked for the duplicates from the 35000 entries and also from itself (2000 entries) and delete it if it is a duplicate and no duplicating operation should be performed on the 35000 entries old data.

I have found some codes but they are deleting even the duplicates of the 35000 entries. I have set the range but even though its not working. Can anyone help me with the best code that takes less time?please thank you

Updating my question with the sample code I have

   A       B            F       G        H       I              Y          
  PTY   39868.5         4       2       540      3      PTY39868.5425403 
  GTY   34446.1234      2       1       230      1      GTY34446.1234212301
  PTY   3945.678                2                2       PTY3945.67822
  GTY   34446.1234      2       1       230      1      GTY34446.1234212301
                  let us say these are old 35000 entries

Explaination to the above example.

The above are the 35000 entries. I have to check A,B,F,G,H,I columns for the dupes, if they are same I have to delete the row, I should not bother about the other columns c,d etc. so what I did is I have used one unused column Y and concatenated these 6 columns values into 1 at Y column using these

  = A2 & B2 & F2 & G2 & H2 &I2 with the respective columns

Now checking the Y column for dupes and delete the entire row. as 2003 supports only for one column as far to my knowledge.

Notice that even the 35000 entries may have duplicates in it but I should not delete them. Example you can see the 2 and last row in my example code are dupes but I should not delete as it is the old data.

   A       B            F       G        H       I              Y          
  PTY   39868.5         4       2       540      3      PTY39868.5425403     'old 
  GTY   34446.1234      2       1       230      1      GTY34446.1234212301   'old
  PTY   3945.678                2                2       PTY3945.67822        'old
  GTY   34446.1234      2       1       230      1      GTY34446.1234212301    'old
  PTY    3945.678       1       1       230      2      PTY3945.678112302      'new
  PTY    39868.5        4       2       540      3      PTY39868.5425403       'new 
  PTY    3945.678       1       1       230      2      PTY3945.678112302      'new

Now note that New entry PTY (from last 2nd) is a duplicate of the original record(PTY at first) So I hava to delete it.And the last new entry is a duplicate of the new entry itself so I should delete it even that . SO in the above code I have to delete only the last 2 rows which are dupes of original record and also from it . But should not delete the GTY which is the dupe but which is in orginal record.

I think I gave a clear view now. Is concatenating them into one cell . Is it better way to approach? as conactenatin for 40000 entries taking just 2 seconds i think that doesnt matter but any more algorithms to these is much aprreciated

I heard counif treats 45.00 and 45.00000 as different is that right may be that was the problem with it? since I have decimal points in my data. I think I should do

    = I2 & H2 & G2 & F2 & A2 & B2

which is better to concatenate? is this or the other i posted before?

Answer 1

This is also a response to some of the comments and solutions made by other members so sorry if it does not straight away answer your question.

Firstly I believe that using excel in a database scenario that raw data and presentation data should be separated. This usually means a single worksheet with raw data and multiple other worksheets with presentation data. Then delete the raw data when necessary or archive.

When speed testing it is very difficult to get a level playing field in excel as there are many things that affect the results. Computer specs, available RAM etc.. Code must first be compiled before running any of the procedures. The test data is also important, when considering duplicates - how many duplicates vs how many rows. This sub loads some test data, altering the amount of rows vs the range of random numbers (duplicates) will give very different results for your code. I don't know what your data looks like so we are kind of working blind and your results may be very different.

'// This is still not very good test data, but should suffice for this situation.
Sub TestFill()
    '// 300000 rows
    For i = 1 To 300000
        '// This populates a random number between 1 & 10000 - adjust to suit
        Cells(i, "A").value = Int((100000 + 1) * Rnd + 1)
    Next
End Sub

If we are talking about advanced filter vs an array & dictonary method then advanced filter will be quicker with a lower amount of rows but once you get above a certain amount of rows then the array method will be quicker. Then see what happens when you change the amount of duplicates.... :) As a guideline or as a general rule using excels built in functions will be faster and I recommend always develop attempting to use these inbuilt functions, however there are often exceptions, like above when removing duplicates. :)

Deleting rows can be slow when looping if used incorrectly. If looping is used then it is important to keep synchronisation between code and the workbook out of the loop. This usually means read data to an array, loop through the data, then load the data from the array back to the presentation worksheet essentially deleting the unwanted data.

Sub RemoveDuplicatesA()

    '// Copy raw data to presentation sheet
    Range("A1", Cells(Rows.Count, "A").End(xlUp)).AdvancedFilter _
        Action:=xlFilterCopy, CopyToRange:=Sheet2.Range("B1"), Unique:=True

End Sub

This will be the fastest method:

Sub RemoveDuplicatesB()        
    Dim vData As Variant, vArray As Variant
    Dim lCnt As Long, lRow As Long

    vData = ActiveSheet.UsedRange.Columns(1).value
    ReDim vArray(0 To UBound(vData, 1), 0)
    lCnt = 0

    With CreateObject("Scripting.Dictionary")
        For lRow = 1 To UBound(vData, 1)
            If Not .Exists(vData(lRow, 1)) Then
                vArray(lCnt, 0) = vData(lRow, 1): lCnt = lCnt + 1
                .Add vData(lRow, 1), Nothing
            End If
        Next lRow
    End With

    '// Copy raw data to presentation sheet
    Sheet2.Range("B1").Resize(lCnt).value = vArray

End Sub

Application transpose has a limitation of 65536 rows but as you are using 2003 you should be fine using it, therefore you can simplify the above code with:

Sub RemoveDuplicatesC()
    Dim vData As Variant
    Dim lRow As Long

    vData = ActiveSheet.UsedRange.Columns(1).value

    With CreateObject("Scripting.Dictionary")
        For lRow = 1 To UBound(vData, 1)
            If Not .exists(vData(lRow, 1)) Then
                .Add vData(lRow, 1), Nothing
            End If
        Next lRow

        '// Copy raw data to presentation sheet or replace raw data
        Sheet2.Columns(2).ClearContents
        Sheet2.Columns(2).Resize(.Count).value = Application.Transpose(.keys)
    End With

End Sub 

EDIT

Okay so @Issun has mentioned you want the entire row deleted. My suggestion was to improve your spreadsheet layout by having a raw data and presentation sheet which means you dont need to delete anything hence it would have been the fastest method. If you dont want to do that and would like to edit the raw data directly then try this:

 Sub RemoveDuplicatesD()
    Dim vData As Variant, vArray As Variant
    Dim lRow As Long       

    vData = ActiveSheet.UsedRange.Columns(1).value
    ReDim vArray(1 To UBound(vData, 1), 0)     

    With CreateObject("Scripting.Dictionary")
        For lRow = 1 To UBound(vData, 1)
            If Not .exists(vData(lRow, 1)) Then
                varray(lRow, 0) = "x"
                .Add vData(lRow, 1), Nothing
            End If
        Next lRow
    End With

    Application.ScreenUpdating = False

    '// Modify the raw data
    With ActiveSheet
        .Columns(2).Insert
        .Range("B1").Resize(lRow).value = vArray
        .Columns(2).SpecialCells(xlCellTypeBlanks).EntireRow.Delete
        .Columns(2).Delete
    End With

    Application.ScreenUpdating = True
End Sub

Answer 2

BIG UPDATE:

It think the original questions threw me off - there may be a problem with the logic in the question. The following assumes you want to delete the cell, not entire row, for the duplicate entries.

• If the 35000 old records do not include duplicates, then all you need to do is remove all duplicates from the entire column - so long as you start from row 1, you run no risk of deleting any of the 'old' rows since no duplicates exist in them.

Here is one way:

Sub UniqueList()

Application.ScreenUpdating = False
Dim vArray As Variant
Dim i As Long, j As Long, lastrow As Long
Dim dictionary As Object
Set dictionary = CreateObject("scripting.dictionary")

lastrow = Range("A" & Rows.Count).End(xlUp).Row
vArray = Range("A1:A" & lastrow).Value

On Error Resume Next
For i = 1 To UBound(vArray, 1)
    For j = 1 To UBound(vArray, 2)
        If Len(vArray(i, j)) <> 0 Then
            dictionary(vArray(i, j)) = 1
        End If
    Next
Next

Columns("A:A").ClearContents
Range("A1").Resize(dictionary.Count).Value = _
Application.Transpose(dictionary.keys)

Application.ScreenUpdating = True

End Sub

• If for some odd reason the 35000 old records DO include dupes and you only want to allow these 35000 records to do so, then you can use 2 dictionaries, but this would be an unusual case since you'd be treating the old records differently than new...

Sub RemoveNewDupes()

Application.ScreenUpdating = False
Dim lastRow As Long
Dim varray As Variant
Dim oldDict As Object, newDict As Object
Set oldDict = CreateObject("scripting.dictionary")
Set newDict = CreateObject("scripting.dictionary")

On Error Resume Next
lastRow = Range("A" & Rows.Count).End(xlUp).Row

'Add old entries to dictionary
varray = Range("A1:A35000").Value
For i = 1 To UBound(varray, 1)
    oldDict.Add varray(i, 1), 1
Next

'Check for dupes
varray = Range("A35001:A" & lastRow).Value
For i = 1 To UBound(varray, 1)
    If oldDict.exists(varray(i, 1)) = False Then
        newDict.Add varray(i, 1), 1
    End If
Next

'Delete and slap back on the unique list
Range("A35001", "A" & Rows.Count).ClearContents
Range("A35001").Resize(newDict.Count).Value = _
Application.Transpose(newDict.keys)

Application.ScreenUpdating = True
End Sub

Thanks to Reafidy for the advice and getting me to relook at this.

Answer 3

Before starting again from scratch your whole code, here are a few things you can try:

Optimize your VBA There are several tips on the web about optimizing vba. In particular, you can do:

'turn off some Excel functionality so your code runs faster
'these two are especially very efficient
Application.ScreenUpdating = False
Application.Calculation = xlCalculationManual
'use these if you really need to
Application.DisplayStatusBar = False
Application.EnableEvents = False

'code goes here

'at the end, restore the default behavior
'calculate the formulas
Application.Calculate
Application.ScreenUpdating = True
Application.Calculation = xlCalculationAutomatic
Application.DisplayStatusBar = True
Application.EnableEvents = True

See here for more information

Optimize your algorithm Especially when your inserting your COUNTIF formula, you can try to fill in instead of inserting the formula in each row.

On the deleting row part, you should try the solution I gave you in your previous thread: Delete duplicate entries in a column in excel 2003 vba to filter first on the True values and then to delete the visible cells. It is probably the fastest way.

[EDIT] Seems like Doc Brown's answer would be probably the best way to handle this (hey, this is a dictionary solution that wasn't written by Issun :)). Anyway, the VBA optimization tips are still relevant because this is quite a slow language.

Answer 4

OK, here's the advancedfilter method. Don't know if it is faster than the dictionary method. It would be interesting to know though, so let me know after you try it. I also included the delete portion so you would have to stop that portion if you want to do a true comparison. Also, you can make this a function instead of a sub and put in your variables, however you want to change it.

Sub DeleteRepeats()

    Dim d1 As Double
    Dim r1 As Range, rKeepers As Range
    Dim wks As Worksheet


    d1 = Timer
    Set wks = ActiveSheet
    Application.EnableEvents = False
    Application.ScreenUpdating = False

    'Make sure all rows are visible
    On Error Resume Next
    wks.ShowAllData
    wks.UsedRange.Rows.Hidden = False
    wks.UsedRange.Columns.Hidden = False
    On Error GoTo 0

    'Get concerned range
    Set r1 = wks.Range("A1:A35000")
    'Filter
    r1.AdvancedFilter Action:=xlFilterInPlace, Unique:=True

    'Get range of cells not to be deleted
    Set rKeepers = r1.SpecialCells(xlCellTypeVisible)
    On Error Resume Next
    wks.ShowAllData
    On Error GoTo 0
    rKeepers.EntireRow.Hidden = True

    'Delete all undesirables
    r1.SpecialCells(xlCellTypeVisible).EntireRow.Delete

    'show all rows
    On Error Resume Next
    wks.UsedRange.Rows.Hidden = False
    On Error GoTo 0

    Application.EnableEvents = False
    Application.ScreenUpdating = False

    Debug.Print Timer() - d1

End Sub

OK, here's a take on Doc's and Issun's use of Dictionaries. Before I wasn't convinced but after looking at it and testing it and comparing to advanced filter, I am convinced, dictionaries are better for this application. I don't know why Excel isn't faster on this point since they should be using faster algorithms, it's not the hiding, unhiding of the rows since that happens very quickly. So if anyone knows, let me know. This procedure takes just over 1 second on my slow computer:

Sub FindDupesAndDelete()

    Dim d1 As Double
    Dim dict As Object
    Dim sh As Worksheet
    Dim v1 As Variant
'    Dim s1() As String
    Dim rDelete As Range
    Dim bUnion As Boolean

    d1 = Timer()
    bUnion = False
    Set dict = CreateObject("Scripting.Dictionary")
    Set sh = ActiveSheet
    v1 = Application.Transpose(sh.Range("A1", "A" _
          & sh.Cells.SpecialCells(xlCellTypeLastCell).row))

'    ReDim s1(1 To UBound(v1))

    Dim row As Long, value As String ', newEntry As Boolean
    For row = 1 To sh.Cells.SpecialCells(xlCellTypeLastCell).row
        value = v1(row)

        If dict.Exists(value) Then
'            newEntry = False
            If bUnion Then
                Set rDelete = Union(rDelete, sh.Range("A" & row))
            Else
                Set rDelete = sh.Range("A" & row)
                bUnion = True
            End If
        Else
'            newEntry = True
            dict.Add value, 1
        End If
'        s1(row) = newEntry

    Next
    rDelete.EntireRow.Delete
'    sh.Range("B1", "B" & UBound(v1)) = Application.Transpose(s1)
    Debug.Print Timer() - d1
End Sub

Answer 5

Okay so now we have some more info here is a solution. It should execute almost instantly.

The code works by filling column y with your concatenate formula. It then adds all of column y to a dictionary and using the dictionary marks each row as a duplicate in column z. It then removes all the duplicates found after row 35000. Then finally it clears both column y and column z to remove the redundant data.

Sub RemoveDuplicates()
    Dim vData As Variant, vArray As Variant
    Dim lRow As Long

    '// Get used range of column A (excluding header) and offset to get column y 
    With ActiveSheet.Range("A2", Cells(Rows.Count, "A").End(xlUp)).Offset(, 24)
        '// Adds the concatenate formula to the sheet column (y)
        .FormulaR1C1 = "=RC[-24]&RC[-23]&RC[-19]&RC[-18]&RC[-17]&RC[-16]"
        '// Adds the formula results to an array
        vData = .Resize(, 1).value
    End With

    '// Re dimension the array to the correct size 
    ReDim vArray(1 To UBound(vData, 1), 0)

    '// Create a dictionary object using late binding
    With CreateObject("Scripting.Dictionary")
        '// Loop through each row in the array
        For lRow = 1 To UBound(vData, 1)
            '// Check if value exists in the array
            If Not .exists(vData(lRow, 1)) Then
                '// Value does not exist mark as non duplicate.
                vArray(lRow, 0) = "x"
                '//  Add value to dictionary
                .Add vData(lRow, 1), Nothing
            End If
        Next lRow
    End With

    '// Turn off screen updating to speed up code and prevent screen flicker
    Application.ScreenUpdating = False    

    With ActiveSheet
        '// Populate column z with the array
        .Range("Z2").Resize(UBound(vArray, 1)) = vArray
        '// Use error handling as speciallcells throws an error when none exist.
        On Error Resume Next
        '// Delete all blank cells in column z
        .Range("Y35001", .Cells(Rows.Count, "Y").End(xlUp)).Offset(, 1).SpecialCells(xlCellTypeBlanks).EntireRow.Delete
        '// Remove error handling
        On Error GoTo 0
        '// Clear columns y and z
        .Columns(25).Resize(, 2).ClearContents
    End With

   '// Turn screen updating back on.
   Application.ScreenUpdating = True
End Sub

NOTE: you can change all references "activesheet" to your sheet codename if you want.

NOTE2: it assumes you have headers and has left row 1 alone.

I have used your columns and test data as best I can. Here is the test fill I used:

Sub TestFill()

    For i = 1 To 37000
        With Range("A" & i)
            .value = Choose(Int(2 * Rnd + 1), "PTY", "GTY")
            .Offset(, 1).value = Round((40000 * (Rnd + 1)), Choose(Int(4 * Rnd + 1), 1, 2, 3, 4))
            .Offset(, 5).value = Int(4 * Rnd + 1)
            .Offset(, 6).value = Int(2 * Rnd + 1)
            .Offset(, 7).value = Choose(Int(2 * Rnd + 1), "230", "540")
            .Offset(, 8).value = Int(3 * Rnd + 1)
        End With
    Next i

End Sub

Answer 6

Lets say you have your entries in column A, and you want the result of your formula in column B (but much faster). This VBA macro should do the trick:

Option Explicit
Sub FindDupes()
    Dim dict As Object
    Dim sh As Worksheet
    Set dict = CreateObject("Scripting.Dictionary")
    Set sh = ActiveSheet

    Dim row As Long, value As String
    For row = 1 To sh.Cells.SpecialCells(xlCellTypeLastCell).row
        value = sh.Range("A" & row).Text
        If dict.Exists(value) Then
            sh.Range("B" & row) = "False"
        Else
            sh.Range("B" & row) = "True"
            dict.Add value, 1
        End If
    Next
End Sub

(Using a dictionary gives here almost linear running time, which should be a matter of seconds for 35.0000 rows, where your original formula had quadratic running time complexity).

Edit: due to your comment: you will have to fill the dictionary first by reading each entry at least once, that is something you cannot avoid easily. What you can avoid is to fill the rows of column B again when they are already filled:

Option Explicit
Sub FindDupes()
    Dim dict As Object
    Dim sh As Worksheet
    Set dict = CreateObject("Scripting.Dictionary")
    Set sh = ActiveSheet

    Dim row As Long, value As String, newEntry As Boolean
    For row = 1 To sh.Cells.SpecialCells(xlCellTypeLastCell).row
        value = sh.Range("A" & row).Text

        If dict.Exists(value) Then
            newEntry = False
        Else
            newEntry = True
            dict.Add value, 1
        End If
        If Trim(sh.Range("B" & row)) = "" Then sh.Range("B" & row) = newEntry
    Next
End Sub

But I suspect this won't be much faster than my first solution.

Answer 7

Now that you have updated that you want the entire rows deleted and that the first 35000 rows are allowed to have dupes, here is a function that will do that for you. I think I came up with a clever method and it's blazing fast, too:

Sub RemoveNewDupes()

Application.ScreenUpdating = False
Dim lastRow As Long
Dim varray As Variant
Dim oldDict As Object, newDict As Object
Set oldDict = CreateObject("scripting.dictionary")
Set newDict = CreateObject("scripting.dictionary")

On Error Resume Next
lastRow = Range("A" & Rows.Count).End(xlUp).Row

'Add old entries to dictionary
varray = Range("A1:A35000").Value
For i = 1 To UBound(varray, 1)
    oldDict.Add varray(i, 1), 1
Next

'Check for dupes
varray = Range("A35001:A" & lastRow).Value
For i = 35000 + UBound(varray, 1) To 35001 Step -1
    If oldDict.exists(varray(i - 35000, 1)) = True Or _
       newDict.exists(varray(i - 35000, 1)) = True Then
        Range("A" & i).EntireRow.Delete
    Else
        newDict.Add varray(i - 35000, 1), 1
    End If
Next

Application.ScreenUpdating = True

'A status message at the end for finishing touch
MsgBox UBound(varray, 1) - newDict.Count & _
" duplicate row(s) found and deleted."

End Sub

How it works:

First I store the 35000 cells into a dictionary file. Then I take a variant array of every cell 35001 onward and loop through them backwards to see if it's in the 35k dictionary or not, or that we haven't come across the value yet in the loop. If it finds that it's a dupe, it deletes the row.

The cool (if I may say) way that it does the row deletion is that when you create the varray, for say A35001 - A37000, it stores them as (1, 1) (2, 1) (...). So if you set "i" to the Ubound of the array + 35000 and step back to 35001, you will loop through all the additions backwardsfrom A37000 to A35001. Then when you want to delete the row, "i" is perfectly set to the row number the value was found in, so you can delete it. And since it goes backwards, it does not screw up the loop!