append connection only queries

[ThomasDay]

Each quarter, CMS publishes new financial data for hospitals.  The reports correctly stack up to each other--so it's an ideal append situation.  They are reasonably large (20mm rows per year and three files).  

I just tried to append by duplicating the initial year's connection only query--giving it a new name--connecting to the new source, and clicking "append query".  To my surprise, it made a separate table with dotted line connections to the original table as well as to the tables it is connected to.

OK, that's harder to read than the situation.  It didn't append the data to the same table--it made another table and wired it to the others.  Won't be long before I won't know which way is up!

Is this the Power Bi way?

If so, I'll probably try to merge the .csv files first and then use the original query.

Thanks,

Tom

[ImkeF]

Yes, this is Power BI - as every query will be loaded to the model.

So in order to proceed as you've already suggested, you copy the step(s) that load the data instead of duplicating the whole query.

As long as you stay in one query, the output will alway stay one table that is connected to the model.

Your combine would look like this: Table.Combine({"LastStepOfFirstPull", "LastStepOfSecondPull", "LastStepOf3rdPull",...})

[ThomasDay]

Thank you, this sounds promising!..   (As a disclaimer, I am still plowing thru the definitive DAX guide--M is for Data Monkey is next but it doesn't seem like a "M" language book. ) 

If I understand what you're saying...with the Query shown below as it exists now--I copy everything below the "let", change the name to Source2 =  and then add a Table.Combine statement?  (This is a connection only, if that matters)

let
Source1 = Csv.Document(File.Contents("C:\Users\thomas\Dropbox\FTRatings\FTRatingsData-Models\DataLoadToPrimaryModelFile\HOSP10_2014_RPT.CSV"),[Delimiter=",", Encoding=1252]),
#"Renamed Columns" = Table.RenameColumns(Source1, {{"Column1", "ReptRecNo"}, {"Column2", "CtrlType(2)"}, {"Column3", "ProvdrNo"}, {"Column4", "NPI"}, {"Column5", "ReptStatus(1)"}, {"Column6", "FYBeginDt"}, {"Column7", "FYEndDt"}}),
#"Renamed Columns1" = Table.RenameColumns(#"Renamed Columns",{{"Column8", "HCRISDt"}}),
#"Removed Other Columns" = Table.SelectColumns(#"Renamed Columns1",{"ReptRecNo", "CtrlType(2)", "ProvdrNo", "FYBeginDt", "FYEndDt", "HCRISDt"}),
#"Changed Type1" = Table.TransformColumnTypes(#"Removed Other Columns",{{"ProvdrNo", Int64.Type}, {"FYBeginDt", type date}, {"FYEndDt", type date}, {"HCRISDt", type date}, {"ReptRecNo", Int64.Type}})
in
#"Changed Type1"

Then the Table.Combine?  What's the table name for the combine statement?

Thank you, Tom

Thanks,

Tom

[pqian · Accepted solution]

@ThomasDay actually, making one query per data set is an acceptable solution, you can consider these tables "backing queries" and choose not to load them (right click -> uncheck Loaded to report). You can even organize them into a folder for browsing needs.

Then, you can have one master query that uses Table.Combine() for all these backing queries, and load that one to the report.

This is probably the easist solution as you don't have to write a single line of M code.

Now, if you want to do all of this in one query, you are going to have to write something. Starting from your query:

let
Source1 = Csv.Document(File.Contents("C:\Users\thomas\Dropbox\FTRatings\FTRatingsData-Models\DataLoadToPrimaryModelFile\HOSP10_2014_RPT.CSV"),[Delimiter=",", Encoding=1252]),
#"Renamed Columns" = Table.RenameColumns(Source1, {{"Column1", "ReptRecNo"}, {"Column2", "CtrlType(2)"}, {"Column3", "ProvdrNo"}, {"Column4", "NPI"}, {"Column5", "ReptStatus(1)"}, {"Column6", "FYBeginDt"}, {"Column7", "FYEndDt"}}),
#"Renamed Columns1" = Table.RenameColumns(#"Renamed Columns",{{"Column8", "HCRISDt"}}),
#"Removed Other Columns" = Table.SelectColumns(#"Renamed Columns1",{"ReptRecNo", "CtrlType(2)", "ProvdrNo", "FYBeginDt", "FYEndDt", "HCRISDt"}),
#"Changed Type1" = Table.TransformColumnTypes(#"Removed Other Columns",{{"ProvdrNo", Int64.Type}, {"FYBeginDt", type date}, {"FYEndDt", type date}, {"HCRISDt", type date}, {"ReptRecNo", Int64.Type}})
in
#"Changed Type1"

Let's parameterize the CSV path:

let
 GetData = (path) => let
 Source1 = Csv.Document(File.Contents(path),[Delimiter=",", Encoding=1252]),
 #"Renamed Columns" = Table.RenameColumns(Source1, {{"Column1", "ReptRecNo"}, {"Column2", "CtrlType(2)"}, {"Column3", "ProvdrNo"}, {"Column4", "NPI"}, {"Column5", "ReptStatus(1)"}, {"Column6", "FYBeginDt"}, {"Column7", "FYEndDt"}}),
 #"Renamed Columns1" = Table.RenameColumns(#"Renamed Columns",{{"Column8", "HCRISDt"}}),
 #"Removed Other Columns" = Table.SelectColumns(#"Renamed Columns1",{"ReptRecNo", "CtrlType(2)", "ProvdrNo", "FYBeginDt", "FYEndDt", "HCRISDt"}),
 #"Changed Type1" = Table.TransformColumnTypes(#"Removed Other Columns",{{"ProvdrNo", Int64.Type}, {"FYBeginDt", type date}, {"FYEndDt", type date}, {"HCRISDt", type date}, {"ReptRecNo", Int64.Type}})
 in
 #"Changed Type1",
 FY14 = GetData("C:\Users\thomas\Dropbox\FTRatings\FTRatingsData-Models\DataLoadToPrimaryModelFile\HOSP10_2014_RPT.CSV")
in
 FY14

Then add FY15, FY16, etc:

let
  GetData = (path) => let
    Source1 = Csv.Document(File.Contents(path),[Delimiter=",", Encoding=1252]),
    #"Renamed Columns" = Table.RenameColumns(Source1, {{"Column1", "ReptRecNo"}, {"Column2", "CtrlType(2)"}, {"Column3", "ProvdrNo"}, {"Column4", "NPI"}, {"Column5", "ReptStatus(1)"}, {"Column6", "FYBeginDt"}, {"Column7", "FYEndDt"}}),
    #"Renamed Columns1" = Table.RenameColumns(#"Renamed Columns",{{"Column8", "HCRISDt"}}),
    #"Removed Other Columns" = Table.SelectColumns(#"Renamed Columns1",{"ReptRecNo", "CtrlType(2)", "ProvdrNo", "FYBeginDt", "FYEndDt", "HCRISDt"}),
    #"Changed Type1" = Table.TransformColumnTypes(#"Removed Other Columns",{{"ProvdrNo", Int64.Type}, {"FYBeginDt", type date}, {"FYEndDt", type date}, {"HCRISDt", type date}, {"ReptRecNo", Int64.Type}})
  in
    #"Changed Type1",
  FY14 = GetData("C:\Users\thomas\Dropbox\FTRatings\FTRatingsData-Models\DataLoadToPrimaryModelFile\HOSP10_2014_RPT.CSV"),
  FY15 = GetData("C:\Users\thomas\Dropbox\FTRatings\FTRatingsData-Models\DataLoadToPrimaryModelFile\HOSP10_2015_RPT.CSV"),
  FY16 = GetData("C:\Users\thomas\Dropbox\FTRatings\FTRatingsData-Models\DataLoadToPrimaryModelFile\HOSP10_2016_RPT.CSV")
  ...

in
  FY16 

Then combine them into the final result

let
  GetData = (path) => let
    Source1 = Csv.Document(File.Contents(path),[Delimiter=",", Encoding=1252]),
    #"Renamed Columns" = Table.RenameColumns(Source1, {{"Column1", "ReptRecNo"}, {"Column2", "CtrlType(2)"}, {"Column3", "ProvdrNo"}, {"Column4", "NPI"}, {"Column5", "ReptStatus(1)"}, {"Column6", "FYBeginDt"}, {"Column7", "FYEndDt"}}),
    #"Renamed Columns1" = Table.RenameColumns(#"Renamed Columns",{{"Column8", "HCRISDt"}}),
    #"Removed Other Columns" = Table.SelectColumns(#"Renamed Columns1",{"ReptRecNo", "CtrlType(2)", "ProvdrNo", "FYBeginDt", "FYEndDt", "HCRISDt"}),
    #"Changed Type1" = Table.TransformColumnTypes(#"Removed Other Columns",{{"ProvdrNo", Int64.Type}, {"FYBeginDt", type date}, {"FYEndDt", type date}, {"HCRISDt", type date}, {"ReptRecNo", Int64.Type}})
  in
    #"Changed Type1",
  FY14 = GetData("C:\Users\thomas\Dropbox\FTRatings\FTRatingsData-Models\DataLoadToPrimaryModelFile\HOSP10_2014_RPT.CSV"),
  FY15 = GetData("C:\Users\thomas\Dropbox\FTRatings\FTRatingsData-Models\DataLoadToPrimaryModelFile\HOSP10_2015_RPT.CSV"),
  FY16 = GetData("C:\Users\thomas\Dropbox\FTRatings\FTRatingsData-Models\DataLoadToPrimaryModelFile\HOSP10_2016_RPT.CSV")
  Combine = Table.Combine({ FY14, FY15, FY16, ... })
in
  Combine

You can even dynamically generate the FY** list and have it respond to a file system change, and automatically pick up new CSV files in a folder.

(ps., i haven't tested the queries above, so there maybe syntax errors, but you get the idea)

Regards,

PQ

[Kane]

Hi, @pqian You provided a very good solution to import multiple data set in one query. I like this style very much. Unfortunately, I met a refresh problem when I published my report which used your method to the Power BI online. When I tried to setup a refresh schedule, it showed that the data set cannot be refreshed automatically since one or more data source are not supported. Did you ever meet this problem ? And do you have any idea to solve this problem ? ------------------------ Here is a simple example. My problem can be reproduced with a report which has only the following query. let GetData = (Path) => let Source = Web.Page(Web.Contents(Path)), Data0 = Source{0��}[Data], #"Changed Type" = Table.TransformColumnTypes(Data0,{{"Header", type text��}, {"Overall rank", Int64.Type��}, {"State", type text��}, {"Cost of living", Int64.Type��}, {"Crime rate", Int64.Type��}, {"Community well-being", Int64.Type��}, {"Health care quality", type text��}, {"Tax rate", Int64.Type��}, {"Weather", Int64.Type��}��}) in #"Changed Type", Web0 = GetData("http://www.bankrate.com/finance/retirement/best-places-retire-how-state-ranks.aspx"), Web1 = GetData("http://www.bankrate.com/finance/retirement/best-places-retire-how-state-ranks.aspx"), Weekly = Table.Combine({Web0, Web1��}) in Weekly Thank you ! Kane

[ThomasDay]

I'll be interested to hear as well, though all of my queries are stable and have handled new files I've put in the folders.  Seems like it must be related to the data source though since the code works well with putting new, identically configured files into a folder.  Is that possible?

[Kane]

Hi, @ThomasDay

With @pqian's method (the nested "let-in" structure), I can easily get data from different files located in and append them to one query, based on which I can create my measures, visuals and reports.

I can also refresh the data in PowerBI Desktop, everything works well. However, after published my report to the PowerBI online, I cannot schedule a refresh for the dataset online.

The  problem can be solved by removing the nested "let-in" structure.

I am puzzled on this. I like @pqian's method, but it seems that it will lead to a refesh problem. Hope someone can help me on this.

Best Regards,

Kane

[ImkeF]

This doesn't seem to be an issue with the method, but with the datasource.

The code you've provided is trowing 404-error-messages in PBI Desktop here already.

[Kane]

Hello, @ImkeF,

I do not think it is the issue of datasource.

Let's try another datasource: https://en.wikipedia.org/wiki/ISO_3166-1

I tried the following two code style. Both of them work well in PowerBI Desktop (I can refresh the data by cliking the Refresh Button mannually).

Unfortunately, once I published the two codes (reports) to the PowerBI online, it shows that I cannot schedule a refresh for the first one.

here is the information:

Code Style 1:

let
    GetData = (Path) =>
    let
        Source = Web.Page(Web.Contents(Path)),
        Data0 = Source{0}[Data],
        #"Changed Type" = Table.TransformColumnTypes(Data0,{{"English short name (upper/lower case)", type text}, {"Alpha-2 code", type text}, {"Alpha-3 code", type text}, {"Numeric code", Int64.Type}, {"Link to ISO 3166-2 subdivision codes", type text}})
    in
    #"Changed Type",
    Web0 = GetData("https://en.wikipedia.org/wiki/ISO_3166-1"),
    Web1 = GetData("https://en.wikipedia.org/wiki/ISO_3166-1"),
    WebCombine = Table.Combine({Web0, Web1})
in
    WebCombine

Code Style 2:

let
    Source = Web.Page(Web.Contents("https://en.wikipedia.org/wiki/ISO_3166-1")),
    Data0 = Source{0}[Data],
    #"Changed Type" = Table.TransformColumnTypes(Data0,{{"English short name (upper/lower case)", type text}, {"Alpha-2 code", type text}, {"Alpha-3 code", type text}, {"Numeric code", Int64.Type}, {"Link to ISO 3166-2 subdivision codes", type text}})
in
    #"Changed Type"

THANK you !

Best Regards,

Kane

[ImkeF]

Hi @Kane,

seems to be a current limitation actually: https://ideas.powerbi.com/forums/265200-power-bi-ideas/suggestions/10927416-web-contents-should-supp...

Wasn't aware of that..

[Kane]

Thank you @ImkeF !

Your information really help me a lot !

Kane

[Kane]

Hi, @pqian You provided a very good solution to import multiple data set in one query. I like this style very much. Unfortunately, I met a refresh problem when I published my report which used your method to the Power BI online. When I tried to setup a refresh schedule, it showed that the data set cannot be refreshed automatically since one or more data source are not supported. Did you ever meet this problem ? And do you have any idea to solve this problem ? ------------------------ Here is a simple example. My problem can be reproduced with a report which has only the following query. let GetData = (Path) => let Source = Web.Page(Web.Contents(Path)), Data0 = Source{0}[Data], #"Changed Type" = Table.TransformColumnTypes(Data0,{{"Header", type text}, {"Overall rank", Int64.Type}, {"State", type text}, {"Cost of living", Int64.Type}, {"Crime rate", Int64.Type}, {"Community well-being", Int64.Type}, {"Health care quality", type text}, {"Tax rate", Int64.Type}, {"Weather", Int64.Type}}) in #"Changed Type", Web0 = GetData("http://www.bankrate.com/finance/retirement/best-places-retire-how-state-ranks.aspx"), Web1 = GetData("http://www.bankrate.com/finance/retirement/best-places-retire-how-state-ranks.aspx"), Weekly = Table.Combine({Web0, Web1}) in Weekly Thank you ! Kane

[ThomasDay]

@pqian--this looks great.  I'm not at my db tomorrow until mid day, but will then experiment and report back.  (What does the "in" line at the end of the query do if I can ask?)

So there is only one query, the Table.Combine would run and append each of the tables resulting from the GetData parameters into one data model table.  Brilliant...

Thank you...I'm looking forward to giving this a go,

Tom

PS: is there an "M" reference or resource you'd recommend for when I've finished the Definitive DAX Guide.

[pqian]

@ThomasDay

The "in" is the return of the closure. You can actually change this to be

let
  FY14 = ..,
  FY15 = ..,
  ..
in
  Table.Combine({FY14, FY15, ...})

which is effectively:

var FY14, FY15...

return Table.Combine({FY14, F15, ...})

You can find a comprehensive guide to the M language on MSDN

https://msdn.microsoft.com/en-us/library/mt211003.aspx

[ThomasDay]

@pqian This is perfect.  Thank you very much...worked like a champ.  The records appended, connections remained, field names changed correctly, columns data types changed properly (seems the order is different not that I should care).  There's only some cleanup to make the queries do all the things I want to every column---and I'll just hit the refresh and off I go.

Thank you again,

Tom