Loading ...
Sorry, an error occurred while loading the content.

304Re: [yws-searchmarketing] Re: addReportRequestWithAccountAggregation Returns variable XML attributes

Expand Messages
  • Yahoo! Search Marketing Monkey
    Dec 5, 2006
      Hey Don,

      We'll take your thoughts into consideration for improving either the
      docs or API.

      -Y!SM Monkey

      don.frazier wrote:
      > Thank you monkey. Upon reviewing the documentation again, I see the
      > following: "KeywordSummary
      > A generated report that contains a pre-determined set of metrics for
      > each keyword."
      > This implies to me that the attributes, being "pre-determined" are
      > not actually "pre-determined" at all. The attributes that appear
      > are "determined" based on others. For example, roas (and others)
      > only appear when numClicks is not 0.
      > After carefully scrutinizing the documentation, I don't see anywhere
      > that it mentions attributes being "optional" or "conditional" or
      > anything else that indicates "sometimes here, sometimes not" nor any
      > mention of "when it's here and when it's not." That would be useful
      > documentation to have, up front, when I am developing custom code to
      > use the output. As it is, the sandbox never delivered variable
      > output so I incorrectly assumed the format was fixed. And as there
      > was no mention of variability in the output, I again only found the
      > problem once I had some live code to test.
      > Your suggestion is of limited value. I am trying to loosely couple
      > my parsing code to the reports (of which Yahoo/Panama is only 1).
      > You suggestion is I tightly couple my code to your format and create
      > a maintenance situation where I have to respond to every change in
      > the format with a corresponding change in code. In my solution, I am
      > loading the reports into a data base table and then normalizing it to
      > a format where I can do my own reporting. As it stands now, the only
      > thing I have to change when a report changes is the format of my bulk
      > load table. I like having a single change point rather than having
      > to re-compile, re-configure, etc.
      > My request is to deliver the reports as document, with the "pre-
      > determined" attributes. If there is no return, for example, just
      > include roas="0.0". My solution was the reverse of your suggestion:
      > I included a list of columns that I could ignore if present. Not
      > good because I still have to update my config file if/when it
      > changes. But I can usually calculate the variable columns from other
      > sources.
      > Thank you for your time.
      > --- In yws-searchmarketing@yahoogroups.com, "Yahoo! Search Marketing
      > Monkey" <plummerm@...> wrote:
      >> Hey Don,
      >> I'd suggest creating a list of all possible columns up front, as
      > they
      >> are documented in the Reports section of the online documentation.
      > Then
      >> just fill in blank/null values for the columns not present in a
      > given
      >> row. You should be able to generalizer this concept to work in all
      > your
      >> xml to tsv conversions.
      >> Hope that helps,
      >> -Y!SM Monkey
      >> don.frazier wrote:
      >>> Hello all ... I am submitting a
      >>> addReportRequestWithAccountAggregation() for my account. In the
      >>> returned XML, there seems to be quite a bit of variance on the
      >>> reports, depending on the the data. In Adcenter and Adwords,
      >>> every "row" in the XML report has the same "columns" regardless
      > of
      >>> the data.
      >>> Panama seems to return different "rows" depending on the actual
      >>> data. For example, the "analytics" element may contain the
      >>> attributes numImpr, numClick, ctr, numConv, revenue, cost and
      >>> averagePosition. But if ctr is anything other than "0.0", then
      > the
      >>> additional columns cpc, numConv, clickConvRate and roas are also
      >>> present.
      >>> My technique for parsing this report is to flatten it into a tab
      >>> delimited file that I then bulk load into my data base for
      > further
      >>> processing. Obviously, the variable columns are killing my
      > process.
      >>> I'd like to keep my XML parser generic cuz it works basically the
      >>> same for any XML data, as long as the data follows a set
      > pattern.
      >>> Any idea if there's a way to get a consistent set of columns in
      > the
      >>> output? Samples below.
      >>> <row cmpgnID="92" adGrpID="100" keywordID="22" urlID="48820"
      >>> accountName="xxx" accountID="30" cmpgnName="s" adGrpName="xx2"
      >>> keywordName="xx" url="http://xxx">
      >>> <analytics numImpr="1" numClick="0" ctr="0.0" numConv="0"
      >>> revenue="0.0" cost="0.0" averagePosition="1.0" />
      >>> </row>
      >>> <row cmpgnID="92" adGrpID="100" keywordID="222" urlID="8"
      >>> accountName="xxx" accountID="30" cmpgnName="s" adGrpName="xx"
      >>> keywordName="xx" url="http://xx">
      >>> <analytics numImpr="26" numClick="5" ctr="0.19230769230769232"
      >>> cpc="0.11" numConv="0" clickConvRate="0.0" revenue="0.0"
      > roas="0.0"
      >>> cost="0.55" averagePosition="1.1153846153846154" />
      >>> </row>
      >>> Yahoo! Groups Links
      > Yahoo! Groups Links
    • Show all 6 messages in this topic