need to take the earliest date of each month - f#

Question

I have a dataset with dates shown below. I need to take the earliest date from each month for all years. How would I go about doing this?
[(1/2/2004 12:00:00 AM);(1/5/2004 12:00:00 AM);
(1/6/2004 12:00:00 AM);(1/7/2004 12:00:00 AM);
(1/8/2004 12:00:00 AM);(1/9/2004 12:00:00 AM);
(1/12/2004 12:00:00 AM);(1/13/2004 12:00:00 AM);
(1/14/2004 12:00:00 AM);(1/15/2004 12:00:00 AM);
(1/16/2004 12:00:00 AM);(1/19/2004 12:00:00 AM);
(1/20/2004 12:00:00 AM);(1/21/2004 12:00:00 AM);
(1/22/2004 12:00:00 AM);(1/23/2004 12:00:00 AM);
(1/26/2004 12:00:00 AM);(1/27/2004 12:00:00 AM);
(1/28/2004 12:00:00 AM);(1/29/2004 12:00:00 AM);
(1/30/2004 12:00:00 AM);(2/2/2004 12:00:00 AM)]
Dataset continues on. Too large to paste here.
EDITED:
let data =
Datacsv.GetSample().Rows
|> Seq.map (fun ((yr,mon),(name,price))
For (name,price) on the last row, I get the error:
TermStructure.fsx(33,36): error FS0001: This expression was expected to have type
'CsvProvider<...>.Row'
but here has type
''a * 'b'

Answer 1

You're almost there. But your last line, with (name, price), isn't quite right. When you call Seq.head data, what is data at that point? Answer: a list of rows. So Seq.head will give you one CSV row. You can't match a CSV row to a (name, price) tuple, and that's what the error message is telling you. Since the input is a row, it was expecting a function that takes ((yr, mon), row) and you gave it ((yr, mon), (name, price)).
In your case, I'd probably back up a step and instead of just doing Seq.head data and feeding that into another Seq.map, I'd do it in one operation. I'll show you my suggestions one step at a time. First, here's the code that you wrote, which is getting the error:
let data =
Datacsv.GetSample().Rows
|> Seq.groupBy (fun row -> row.DATE.Year,row.DATE.Month)
|> Seq.map (fun ((yr,mon),data) -> ((yr, mon), Seq.head data))
|> Seq.map (fun ((yr,mon),(name,price))
My first thought is that having two different things named data in your code is going to be confusing. So let's change the variable name in the first Seq.map. The thing you've named data in that function is a sequence of rows, so let's call it rows instead:
let data =
Datacsv.GetSample().Rows
|> Seq.groupBy (fun row -> row.DATE.Year,row.DATE.Month)
|> Seq.map (fun ((yr,mon),rows) -> ((yr, mon), Seq.head rows))
|> Seq.map (fun ((yr,mon),(name,price))
Now we'll fix the second Seq.map by removing it and merging its operation into the first Seq.map, as follows:
let data =
Datacsv.GetSample().Rows
|> Seq.groupBy (fun row -> row.DATE.Year,row.DATE.Month)
|> Seq.map (fun ((yr,mon),rows) ->
let row = Seq.head rows
((yr, mon), (row.NAME, row.PRICE)))
Note that I'm assuming that since your CSV file has a column called DATE in all caps (which has translated to a DATE property on your row objects in F#), it also contains NAME and PRICE columns in all caps too. If those columns are named something else, adjust the row.NAME and row.PRICE parts of the code accordingly.