I have a dataframe with columns: A(continuous variable) and B(discrete 1 or 0). The df is initially sorted by A variable.
I need to order the dataframe so for each set of X rows, there are Y rows with value 1 in B column, and (X-Y) rows with 0 (B column) (when possible!). But these sets should have variable A in desceding order. X and Y are input by the user
Example:
X=4, Y=3
Rows 0-11 are ok, since the sets (0-3),(4-7) and (8-11) has 3 rows with 1 in column B and only one row with 0 AND variable A is descending. However, rows 12-15 are not ok, since there are 2 rows with 1(variable B) and two with 0. Row 17 would replace row 15 to make this set valid. There is no problem if the last rows has 0 in variable B, since there isn't any with value 1.
The code should be general enough to run on dataframes with different number of rows.
Any ideas?
I am new to GAMS and I have a table data which has 3 rows and 6 columns. I want to pull each row and use its data for two parameters(pull each row which has 6 element and use the first three elements for one parameter and the other three elements for the second parameter) using loop or for statement. i tried to use both of them but for the loop i received zero value for my parameter which is incorrect and for the for statement i received some errors.
this is my code for the first row which both 'loop' and 'for' are used (i used them separately each time but for show what was my code i just wrote them together).
Please help me.
Thanks
scalars j;
sets
o /red,green,blue/
p /b1,b2,b3,p1,p2,p3/
k /1*3/;
Table sup(*,*)
b1 b2 b3 p1 p2 p3
red 12 15 20 200 50 50
green 16 17 0 150 50 0
blue 13 18 0 100 50 0 ;
parameters Bid_Red(k),Pmax_Red(k),t;
*for statement***************
for(j= 1 to 3,
t=card(o)+j;
Bid_Red(k)$( ord(k) = j )=sup('red',j);
Pmax_Red(k)$( ord(k) = j )=sup('red',t);
);
*loop statement***************
t=card(o);
loop(k,
Bid_Red(k)=sup('red',k);
Pmax_Red(k)=sup('red',k+t);
);
display Bid_red, Pmax_Red
One of the core features of GAMS is how it deals with set structures and indexing. I'd recommend looking at the excellent documentation, for example on set definition https://www.gams.com/latest/docs/UG_SetDefinition.html, to really get a feel for how to get the best out of it.
In your case, you can proceed as follows. p is a set. Create some subsets of it p_ and b_, given by the syntax subset_name(set_name).
sets p_(p) / p1, p2, p3 /,
b_(p) / b1, b2, b3 /;
Create parameters over appropriate dimensions (i.e. the full set), and define them over the subset you are interested in:
parameters bid_red(o,p),pmax_red(o,p);
bid_red(o,b_) = sup(o,b_);
pmax_red(o,p_) = sup(o,p_);
Then display bid_red, pmax_red; gives:
---- 21 PARAMETER bid_red
b1 b2 b3
red 12.000 15.000 20.000
green 16.000 17.000
blue 13.000 18.000
---- 21 PARAMETER pmax_red
p1 p2 p3
red 200.000 50.000 50.000
green 150.000 50.000
blue 100.000 50.000
If you do want to select individual rows, you can use e.g. pmax_red('red',p_) in your code. This is essentially just a special case of subsetting in which the subset is of size 1.
Let's say I have two columns.
3.5463 11
4.5592 12
1.6993 111
0.92521 112
1.7331 121
2.1407 122
1.4082 1111
2.0698 1112
2.3973 1121
2.4518 1122
1.1719 1211
1.153 1212
0.67139 1221
0.64744 1222
1.3705 11111
0.9557 11112
0.64868 11121
0.7325 11211
0.58874 11212
0.86673 11221
0.17075 11222
0.64026 12111
0.80229 12112
0.43422 12122
1.0405 12211
0.63376 12212
0.56491 12221
0.34626 12222
0.81631 111111
0.91837 111112
0.70013 111121
0.87384 111122
1.1474 111211
0.47411 111221
0.12249 111222
0.56728 112111
0.88169 112112
0.14509 112121
0.68655 112211
0.36274 112212
1.1652 121111
0.99314 121112
0.42024 121121
0.23937 121122
1.0346 122111
0.64642 122112
0.15632 122121
0.41725 122122
0.40793 122211
In the first column, there is a number. With every one of those numbers, in the second column, is an associated ID. Now, there are some blank rows that do not contain any numbers in them.
Define one of these numbers to be a "daughter" of another number if the ID of the first number is the same as the ID of the second, with an extra digit on the end. For example, both IDs 11211 and 11212 are daughters of 1121, because the ID of 1121 has an extra digit, either a 1 or a 2, added onto the end to form the ID of its daughters. Thus, 1121 is the parent of both 11211 and 11212.
Here is what I want the macro to do. It must output a third column which contains, for every row, a cumulative sum of the number of the first column in that row, plus the parent number of that number, and the parent number of the parent number, etc. all the way up until it reachers either 11 or 12. It will begin by simply outputting the numbers in column 1 for 11 and 12 in the third column. Then, in a loop beginning with 111, it will add up the cumulative sum of every row (the number in that row plus the third column output of the parent), only if that row has a number and an id, and only if the parent exists and has an output in column 3. So for example, the number in the 3rd column of the row with ID 11222 should be the number in column 1 of that row, plus that of 1122, plus that of 112, plus that of 11. So, 0.17075+2.4518+0.92521+3.5463, or 7.09406. However, if you try to do this for ID 111221, you will notice that the row where the parent 11122 should be is empty. Thus, the parent does not exist, and no value will be outputted in column 3 for 111221.
I would greatly appreciate it if someone has some time on their hands to code up this VBA macro for me in exchange for an accepted solution.
Thanks
I don't think a macro is needed, just some formulas. First, I put a header on my columns of data, such as "value," and "id." If you then highlight the column labels (i.e., A and B) and sort by B ("id") then A ("value"), you'll group your blank rows. You can then delete those rows. Now you have the data almost ready. When I did this, I converted the id column to text, as opposed to a number value, so if I sort the table by id, the pattern will be, "11, 111, 1111," and so on, instead of, "11, 12, 111, 112, 121." Then, I added columns to separate the separate characters or levels of the ids. This is to help with parents and children. You can use text-to-columns, or a MID formula, but what I did was have 6 more columns to the right. For each id row, each column would either have a "1," a "2," or a blank (null) value. Then I added another column, calling it "level." I used a formula like COUNTA across all my id splitting columns. So, for 11, my level value was 2. 111 would be 3, 11221 would be 5, and so on. This gives me the id level (parent, child, grandchild, etc). Then I added my final column to the right to compute my cumulative sum of the values. In concept I have one big nested IF statement, but in practice, I needed two. My formula says, if the row above me has a lower level number (i.e., it is some kind of parent), add the value of the current row to the value of the above row. Otherwise, keep going up a row till I do get a parent, and add the current row value to that number.
My final formula for all but the first 5 rows of data was (in the 6th row of data):
=if(K6
rest of answer is below
=if(K6<K7,L6+C7,if(K5<K7,L5+C7,if(K4<K7,L4+C7,if(K3<K7,L3+C7,if(K2<K7,L2+C7,C7)))))
The values were column C, the original id in column D, the id split columns were E through J, the level column was K, and my formula was in L. This formula can be copied down the table. For the first 4 rows, you just need 1 less IF statement each row you go up. The fifth row of data might take the above formula; it depends how it will deal with the column headers in row one. The formula on the 4 row of data might be:
=if(K4<K5,L4+C5,if(K3<K5,L3+C5,if(K2<K5,L2+C5,if(K1<K5,L1+C5,C5))))
I'm still learning how to format these comments, so I'll try to provide a sample of the layout I have...
C D E F G H I J K L
1 value id 1 2 3 4 5 6 lvl cumul_sum
2 3.546300 11 1 1 2 3.546300
3 1.699300 111 1 1 1 3 5.245600
4 1.408200 1111 1 1 1 1 4 6.653800
5 1.370500 11111 1 1 1 1 1 5 8.024300
6 0.816310 111111 1 1 1 1 1 1 6 8.840610
7 0.918370 111112 1 1 1 1 1 2 6 8.942670
8 0.955700 11112 1 1 1 1 2 5 7.609500
So for example, the number in the 3rd column of the row with ID 11222 should be the number in column 1 of that row, plus that of 1122, plus that of 112, plus that of 11. So, 0.17075+2.4518+0.92521+3.5463, or 7.09406.However, if you try to do this for ID 111221, you will notice that the row where the parent 11122 should be is empty. Thus, the parent does not exist, and no value will be outputted in column 3 for 111221.
As a native worksheet array formula¹ in D1,
=IF(LEN(B1), SUM(SUMIFS(A$1:INDEX(A:A, MATCH(1E+99, A:A)),
B$1:INDEX(B:B, MATCH(1E+99, A:A)), LEFT(B1, ROW(INDIRECT("2:"&LEN(B1)))))), TEXT(,))
The above does not compensate for missing parents (null string). It totals everything it can find and uses zero for missing parents.
As a VBA UDF² in E1,
Function conditionalCumulativeSum(nums As Range, _
ids As Range, sib As Range, _
Optional nullOnBlank As Boolean = True)
Dim i As Integer
'truncate any full column reference to the UsedRange
Set nums = Intersect(nums, nums.Parent.UsedRange)
'match the nums and ids ranges
Set ids = ids.Resize(nums.Rows.Count, nums.Columns.Count)
For i = Len(sib.Value2) To 2 Step -1
If nullOnBlank And IsError(Application.Match(--Left(sib, i), ids, 0)) Then
conditionalCumulativeSum = vbNullString
Exit For
End If
conditionalCumulativeSum = conditionalCumulativeSum + _
Application.SumIfs(nums, ids, Left(sib, i))
Next i
If i = 0 Then conditionalCumulativeSum = vbNullString
End Function
The above defaults to return a null string when it encounters any missing parent through the hereditary chain. This can be turned off by adding FALSE as the optional fourth parameter and then the UDF will behave identically to the native formula.
Results from sample data
¹ Array formulas need to be finalized with Ctrl+Shift+Enter↵. If entered correctly, Excel with wrap the formula in braces (e.g. { and }). You do not type the braces in yourself. Once entered into the first cell correctly, they can be filled or copied down or right just like any other formula. Try and reduce your full-column references to ranges more closely representing the extents of your actual data. Array formulas chew up calculation cycles logarithmically so it is good practise to narrow the referenced ranges to a minimum. See Guidelines and examples of array formulas for more information.
² A User Defined Function (aka UDF) is placed into a standard module code sheet. Tap Alt+F11 and when the VBE opens, immediately use the pull-down menus to Insert ► Module (Alt+I,M). Paste the function code into the new module code sheet titled something like Book1 - Module1 (Code). Tap Alt+Q to return to your worksheet(s).
I have a table in Qlikview with 2 columns:
A B
a 10
b 45
c 30
d 15
Based on this table, I have a formula with full acumulation defined as:
SUM(a)/SUM(TOTAL a)
As a result,
A B D
b 45 45/100=0.45
c 30 75/100=0.75
d 15 90/100=0.90
a 10 100/100=1
My question is. how do I mark in colour the values in column A that have on column D <=0.8)?
The challenge is that D is defined with full accumulation, but if I reference D in a formula, it doesn't consider the full accumulation!
I tried with defining a formula E=if(D>0.8,'Y','N') but this formula doesn't take the visible (accumulated) value for D unfortunately, instead it takes the D with no accumulation. If this worked, I would have tried to hide (not disable) E and reference it from the dimensions column of the table , Text colour option. Any ideas please?? Thanks
You can't get an expression column's value from within a dimension or it's properties, because the expression columns rely on the dimensions provided. It would create an endless loop. Your options are:
Apply your background colour to the expression columns, not the dimensions. This would actually make more sense as the accumulated values would have the colour, not the dimension.
When loading this specific table, have QlikView create a new column that contains the accumulated values of B. This would mean, however, that the order of your chart-table would need to be fixed for the accumulations to make any sense.
Use aggregation to create a temporary table and accumulate the values using RangeSum(). Note this will only accumulate properly if the table is ordered in Ascending order of Column A
=IF(Aggr(RangeSum(Above(Sum(B),0,10)),A)/100>0.8,
rgb(0,0,0),
rgb(255,0,0)
)
I have over 3000 values in column A. I have a value of x in column B. I want excel to look through the values in column A and give back a "yes" if there is a value bigger than x, yet smaller than x+7 (x+7>value to be found in column A >x). If such value(s) does not exist, then display "no"
Here's an example:
Column A
2: 11.2
3: 11.3
4: 11.4
5: 13.5
6: 13.6
7: 20.5
8: 20.6
9: 30.5
Column B
2: 11.1
3: 20.7
In this case, since there are values in Column 1 that are bigger than 11.1 and within the range (smaller than B2+7, and bigger than B2), I need excel to give back "yes". If possible, it would be ideal to also give the first value after the specific value in column B.
Here's what I have tried so far but have had no success:
=IF(AND((B2+7)>A1:A3000>B2),"yes","no")
=IF(AND((B2+7)>$A$2:$A$3000,$A$2:$A$3000>B2),"yes","no")
How can I do this in Excel? is there a way to do this other than using IF?
Forgive me if I am not understanding the question, but isn't the answer:
=IF(AND((B2+7)>$A$2:$A$10,$A$2:$A$10>B2),"yes","no")
Where that would be the equation in C2, testing B2 for entries in the list that spans A2 - A10. You'd copy that equation down the column for all the entries in B column.
Try this (for cell B2)
=IF(SUMPRODUCT((A:A>B2)*(A:A<B2+7)),"Yes","No")
and copy down as far as required.
For the second part, to return the next larger value, try this
=IF(SUMPRODUCT((A:A>B2)*(A:A<B2+7)),INDEX(A:A,MATCH(B2,A:A,1)+1),"No")
Note that this requires the data in column A to be sorted ascending (as your sample ata is) and for cell A to contain '0'. If these conditions are not possible, then you might have to consider a VBA user defined function.
I don't understand what you require but maybe =IF(AND(A2>B$2,A2<B$2+7),"yes","no") copied down would serve to test B2 and B2+7 against each of the Column A values.