March 31 - April 2, 2025, in Las Vegas, Nevada. Use code MSCUST for a $150 discount! Early bird discount ends December 31.
Register NowBe one of the first to start using Fabric Databases. View on-demand sessions with database experts and the Microsoft product team to learn just how easy it is to get started. Watch now
@dufoq3 Need help pls ! Continuing from a previous post: Solved: Re: End date based on next row (Index+1) for each ... - Microsoft Fabric Community
I am looking to calculate the End date. We need to look at the combination of (MainAccountID, BU and PV) group and if a transaction occurs for the same month, then next row start date minus 1. The dates have to be a business day. In the first record,
Main Account ID, BU and PV (34110-100-3) has a subsequent transaction on 2018-07-09. The previous record end date is equal to previous business day 2018-07-06.
StartDate | MainAccountID | BU | PV | Transaction | EndDate |
2018-07-03 | 34110 | 100 | 3 | 5 | 2018-07-06 |
2018-07-09 | 34110 | 100 | 3 | 6 | 2018-07-12 |
2018-07-13 | 34110 | 100 | 3 | 4 | 2018-07-31 |
2018-07-03 | 34110 | 100 | 4 | 1 | 2018-07-06 |
2018-07-09 | 34110 | 100 | 4 | 2 | 2018-07-31 |
Solved! Go to Solution.
Hi @mthiru,
I've added 1 more row (comfirm if this is expected result or not).
If there is something wrong, we need to consider sort order. Let me know.
Result:
let
fnShift = (tbl as table, col as text, shift as nullable number, optional newColName as text) as table =>
//v 3. parametri zadaj zaporne cislo ak chces posunut riadky hore, kladne ak dole, 4. je nepovinny (novy nazov stlpca)
let
a = Table.Column(tbl, col),
b = if shift = 0 or shift = null then a else if shift > 0
then List.Repeat({null}, shift) & List.RemoveLastN(a, shift)
else List.RemoveFirstN(a, shift * -1) & List.Repeat({null}, shift * -1),
c = Table.FromColumns(Table.ToColumns(tbl) & {b}, Table.ColumnNames(tbl) &
( if newColName <> null then {newColName} else
if shift = 0 then {col & "_Duplicate"} else
if shift > 0 then {col & "_PrevtValue"}
else {col & "_NextValue"} ))
in
c,
Source = Table.FromRows(Json.Document(Binary.Decompress(Binary.FromText("i45WMjIwtNA1MNc1MFbSUTI2MTQ0ANKGBiASJGKqFKuDpMgSqyIzFEWG2E0yQTUJU5EJiEXIOpAiIyRFlrqGhlgVGSvFxgIA", BinaryEncoding.Base64), Compression.Deflate)), let _t = ((type nullable text) meta [Serialized.Text = true]) in type table [StartDate = _t, MainAccountID = _t, BU = _t, PV = _t, Transaction = _t]),
ChangedType = Table.TransformColumnTypes(Source,{{"StartDate", type date}}),
Ad_PrevWorkingDate = Table.AddColumn(ChangedType, "Prev. Working Date", each List.Max(List.Select(List.Dates(Date.AddDays([StartDate], -1), 3, #duration(-1,0,0,0)), (x)=> Date.DayOfWeek(x, Day.Monday) < 5)), type date),
GroupedRows = Table.Group(Ad_PrevWorkingDate, {"MainAccountID", "BU", "PV"}, {{"All", each
[ a = fnShift(_, "Prev. Working Date", -1 , "PWD Shift"), //Prev. Working Date Shift
b = fnShift(a, "StartDate", -1 , "SD Shift"), //Prev. Start Date Shift
c = Table.AddColumn(b, "EndDate", (x)=> if Date.ToText(x[StartDate], [Format="yyyy-MM"]) = Date.ToText(x[SD Shift], [Format="yyyy-MM"]) then x[PWD Shift] else Date.EndOfMonth(x[StartDate]), type date), //Add EndDate
d = Table.RemoveColumns(c, {"Prev. Working Date", "PWD Shift", "SD Shift"}) //Remove helper columns
][d], type table}}),
CombinedAll = Table.Combine(GroupedRows[All])
in
CombinedAll
@dufoq3 this is actually not what I was expecting. I may try and post another post to get help. Ideally, If I do with this less code and manually with less code intensive, using wizard in power query will be better for me.
If you want to understand the logic of previous/next row value, read this article.
@dufoq3 Thank you!
I'd be no where without your help on this end date solution. If we group on AccountId, BU and PV then this should work. I have query2 with the above code and am displaying the columns in a table. I add filters to EC and PV. The first two lines should be combined, I will look into grouping that and it should work fine.
You're welcome. Here is updated code - matching with new sample:
let
fnShift = (tbl as table, col as text, shift as nullable number, optional newColName as text, optional _type as type) as table =>
//v 3. parametri zadaj zaporne cislo ak chces posunut riadky hore, kladne ak dole, 4. je nepovinny (novy nazov stlpca), 5. je nepovinny typ
let
a = Table.Column(tbl, col),
b = if shift = 0 or shift = null then a else if shift > 0
then List.Repeat({null}, shift) & List.RemoveLastN(a, shift)
else List.RemoveFirstN(a, shift * -1) & List.Repeat({null}, shift * -1),
c = Table.FromColumns(Table.ToColumns(tbl) & {b}, Table.ColumnNames(tbl) &
( if newColName <> null then {newColName} else
if shift = 0 then {col & "_Duplicate"} else
if shift > 0 then {col & "_PrevtValue"}
else {col & "_NextValue"} )),
d = Table.TransformColumnTypes(c, {List.Last(Table.ColumnNames(c)), if _type <> null then _type else type any})
in
d,
Source = Table.FromRows(Json.Document(Binary.Decompress(Binary.FromText("tdI9C4MwEAbgvxIyG7i7fJm9dBA6dRQHLYEulRIU6r+vGopQYsFA3yEhB3lyHKlr3o+Pzgde8MG/hnk7+ZtfKkxiwQiI5hoiCAB0QqAV8bxdQLcsdisIkgoclGh4U/zjgYM2fdn0yy7JOW0UJvmq7cc2TIz0qsvDOsKaJH72XYi6zdQVaTknqV/acLszUrmNmyV2h54+8z4Oa+1wr+dq7D2TkCnLmKR89c8hfpQyV7cG14E0bw==", BinaryEncoding.Base64), Compression.Deflate)), let _t = ((type nullable text) meta [Serialized.Text = true]) in type table [EC = _t, Name = _t, AccountingDate = _t, Ledger = _t, MainAccountId = _t, AcctDescription = _t, BU = _t, PV = _t, Description = _t, Amount = _t]),
ChangedType = Table.TransformColumnTypes(Source,{{"AccountingDate", type date}}),
Ad_PrevWorkingDate = Table.AddColumn(ChangedType, "Prev. Working Date", each List.Max(List.Select(List.Dates(Date.AddDays([AccountingDate], -1), 3, #duration(-1,0,0,0)), (x)=> Date.DayOfWeek(x, Day.Monday) < 5)), type date),
GroupedRows = Table.Group(Ad_PrevWorkingDate, {"MainAccountId", "BU", "PV"}, {{"All", each
[ a = fnShift(_, "Prev. Working Date", -1 , "PWD Shift"), //Prev. Working Date Shift
b = fnShift(a, "AccountingDate", -1 , "SD Shift"), //Prev. Start Date Shift
c = Table.AddColumn(b, "EndDate", (x)=> if Date.ToText(x[AccountingDate], [Format="yyyy-MM"]) = Date.ToText(x[SD Shift], [Format="yyyy-MM"]) then x[PWD Shift] else Date.EndOfMonth(x[AccountingDate]), type date), //Add EndDate
d = Table.RemoveColumns(c, {"Prev. Working Date", "PWD Shift", "SD Shift"}) //Remove helper columns
][d], type table}}),
CombinedAll = Table.Combine(GroupedRows[All])
in
CombinedAll
@dufoq3 The new code resolves the null values. I just noticed the end date sometimes is not as expected. In the first row, we have end date as Dec 30 2022. It should be Dec 31 2022.
EC | Name | AccountingDate | EndDate | Ledger | MainAccountId | AcctDescription | BU | PV | Description | Amount |
number | text | December 31, 2022 | December 30, 2022 | 110-0019--17- | 110 | text | 19 | 17 | text | -234090816 |
number | text | December 31, 2022 | December 31, 2022 | 110-0019--17- | 110 | text | 19 | 17 | text | 90816 |
number | text | December 31, 2022 | December 31, 2022 | 120-0019--17- | 120 | text | 19 | 17 | text | 82995641 |
number | text | January 25, 2023 | January 31, 2023 | 120-0019--17- | 120 | text | 19 | 17 | text | 1000000 |
number | text | February 27, 2023 | February 28, 2023 | 120-0019--17- | 120 | text | 19 | 17 | text | 4253333 |
number | text | March 24, 2023 | March 31, 2023 | 120-0019--17- | 120 | text | 19 | 17 | text | 1666667 |
number | text | May 31, 2023 | May 31, 2023 | 120-0019--17- | 120 | text | 19 | 17 | text | 5591333 |
number | text | June 30, 2023 | June 30, 2023 | 120-0019--17- | 120 | text | 19 | 17 | text | 3333333 |
number | text | September 8, 2023 | September 30, 2023 | 120-0019--17- | 120 | text | 19 | 17 | text | 3761667 |
It is correct from my point of view, because you've asked for:
"if a transaction occurs for the same month, then next row start date minus 1. The dates have to be a business day"
In the first row you have date 31.12.2022 which is in the same month as second row (again 31.12.2022) and you wanted to return last business date before 2nd row date which is 30.12.2022 - Friday.
Hi @mthiru,
I've added 1 more row (comfirm if this is expected result or not).
If there is something wrong, we need to consider sort order. Let me know.
Result:
let
fnShift = (tbl as table, col as text, shift as nullable number, optional newColName as text) as table =>
//v 3. parametri zadaj zaporne cislo ak chces posunut riadky hore, kladne ak dole, 4. je nepovinny (novy nazov stlpca)
let
a = Table.Column(tbl, col),
b = if shift = 0 or shift = null then a else if shift > 0
then List.Repeat({null}, shift) & List.RemoveLastN(a, shift)
else List.RemoveFirstN(a, shift * -1) & List.Repeat({null}, shift * -1),
c = Table.FromColumns(Table.ToColumns(tbl) & {b}, Table.ColumnNames(tbl) &
( if newColName <> null then {newColName} else
if shift = 0 then {col & "_Duplicate"} else
if shift > 0 then {col & "_PrevtValue"}
else {col & "_NextValue"} ))
in
c,
Source = Table.FromRows(Json.Document(Binary.Decompress(Binary.FromText("i45WMjIwtNA1MNc1MFbSUTI2MTQ0ANKGBiASJGKqFKuDpMgSqyIzFEWG2E0yQTUJU5EJiEXIOpAiIyRFlrqGhlgVGSvFxgIA", BinaryEncoding.Base64), Compression.Deflate)), let _t = ((type nullable text) meta [Serialized.Text = true]) in type table [StartDate = _t, MainAccountID = _t, BU = _t, PV = _t, Transaction = _t]),
ChangedType = Table.TransformColumnTypes(Source,{{"StartDate", type date}}),
Ad_PrevWorkingDate = Table.AddColumn(ChangedType, "Prev. Working Date", each List.Max(List.Select(List.Dates(Date.AddDays([StartDate], -1), 3, #duration(-1,0,0,0)), (x)=> Date.DayOfWeek(x, Day.Monday) < 5)), type date),
GroupedRows = Table.Group(Ad_PrevWorkingDate, {"MainAccountID", "BU", "PV"}, {{"All", each
[ a = fnShift(_, "Prev. Working Date", -1 , "PWD Shift"), //Prev. Working Date Shift
b = fnShift(a, "StartDate", -1 , "SD Shift"), //Prev. Start Date Shift
c = Table.AddColumn(b, "EndDate", (x)=> if Date.ToText(x[StartDate], [Format="yyyy-MM"]) = Date.ToText(x[SD Shift], [Format="yyyy-MM"]) then x[PWD Shift] else Date.EndOfMonth(x[StartDate]), type date), //Add EndDate
d = Table.RemoveColumns(c, {"Prev. Working Date", "PWD Shift", "SD Shift"}) //Remove helper columns
][d], type table}}),
CombinedAll = Table.Combine(GroupedRows[All])
in
CombinedAll
Thank you @dufoq3. I have to try it with the new code. Previously I gotten the end date as expected, with one exception where in the data the PV is null. I get a null end date. Sorry, I keep forgetting to include all possible conditions in the data. We always have a MainAccountID and BU value but sometimes the PV is blank.
March 31 - April 2, 2025, in Las Vegas, Nevada. Use code MSCUST for a $150 discount!
Arun Ulag shares exciting details about the Microsoft Fabric Conference 2025, which will be held in Las Vegas, NV.