"MarketBasket Analysis Pivot-Problem"

SunnyLotusFloweSunnyLotusFlowe Member Posts: 37 Contributor II
edited June 2019 in Help
Hi all,

i have a Problem with my workflow:

I have a Data set like

CardID | Product
10150  softdrink
10236  beer
...

Everytime i wanna do pivoting, the Product-Column disapears.
The Pivoting has the following configuration:

group attribute:  cardid
index attribute:  Product

I dont understand why the Product disapears ....

greetings

SunnyLotusFlower






Answers

  • SunnyLotusFloweSunnyLotusFlowe Member Posts: 37 Contributor II
    Does noone got an idea about that problem? is it a bug or sm like that?

    greetings

    Lotus
  • IngoRMIngoRM Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, Community Manager, RMResearcher, Member, University Professor Posts: 1,751 RM Founder
    No, not a bug but you would need an additional column containing the values which should be placed into the resulting table. Try:

    CardID | Product | Amount
    10150  softdrink  1
    10236  beer  1
    ...

    and you will be fine.

    Cheers,
    Ingo
  • SunnyLotusFloweSunnyLotusFlowe Member Posts: 37 Contributor II
    hm didnt work.    i have the following settings for the Pivoting-Operator:

    Group Attribute cardid
    index attribute: product

    and i have created a new column amount

    greetings

    Lotus
    ____________________
    Update
    i noticed that the Pivoting Operator had removed my Product column and doesnt show resulting columns





  • IngoRMIngoRM Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, Community Manager, RMResearcher, Member, University Professor Posts: 1,751 RM Founder
    There is a working example of how to use the Pivoting operator on myexperiment.org. Just install the Community Extension and download the process named "Pivoting". There you will see how this process works.

    Cheers,
    Ingo
  • SunnyLotusFloweSunnyLotusFlowe Member Posts: 37 Contributor II
    i got the workflow. i can replicate the problem, if i remove the add noise operator.

    but why is this 'add noise'-operator there ? that is confusing...


    greetings

    Lotus
  • IngoRMIngoRM Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, Community Manager, RMResearcher, Member, University Professor Posts: 1,751 RM Founder
    Hi,

    you would need an additional column containing the values which should be placed into the resulting table.

    why is this 'add noise'-operator there ? that is confusing...
    Make an educated guess... what could this noise attribute be used for?

    Right: the values which should be placed in the resulting table!

    Cheers,
    Ingo
  • SunnyLotusFloweSunnyLotusFlowe Member Posts: 37 Contributor II
    hm i found something interesting i think at least.


    if i construct a column count, with the instances all = 1 ( all examples have the SAME !!!  entry ) then the pivot-operator doesnt perform his rotating .

    Why i cant do that ??? did i forget a important detail ??




    greetings Lotus
  • SunnyLotusFloweSunnyLotusFlowe Member Posts: 37 Contributor II
    or is this a bug ?

    greetings

    Lotus
  • IngoRMIngoRM Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, Community Manager, RMResearcher, Member, University Professor Posts: 1,751 RM Founder
    Hi,

    thanks for pointing out this issue. This could indeed be a bug. I have filed a bug report at

    http://bugs.rapid-i.com/show_bug.cgi?id=380

    Cheers,
    Ingo
  • SunnyLotusFloweSunnyLotusFlowe Member Posts: 37 Contributor II
    the curious problem is that, this 'feature lead to my problems i described all the time.

    i think i have another bug for u:

    can i send u a data set?

    because not all column are displayed when pivoting...

  • IngoRMIngoRM Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, Community Manager, RMResearcher, Member, University Professor Posts: 1,751 RM Founder
    Hi,

    can i send u a data set?
    no, but you could attach it to the bug or to one of our developers.

    Cheers,
    Ingo
  • CrazyTCrazyT Member Posts: 8 Contributor II
    Bug seems to be same as this one:
    http://bugs.rapid-i.com/show_bug.cgi?id=245
  • SunnyLotusFloweSunnyLotusFlowe Member Posts: 37 Contributor II
    with my data set i constructed the follow Workflow:

    Read Csv -> Generate Empty -> Numerical to polynominal ->

    replace Missing Values -> set data -> pivoting -> set data (2) -> replace Missing

    ______________

    ok to bypass that problem i described before i use the operator 'set data'.
    i change the first example to  2 the rest of the counts got a '1' . and then i rotate
    with the pivoting. if i look in the second Set Data Operator there must be more than 3 column
    which i can choose.

    in other words: the pivoting operator got 15 regular attributes and i in the second set data op i
    only can choose between 3. there must be 15!


    greetings

    Lotus


    _________________

    this i a part of the data. it should be placed into a csv-data- and then the workflow i wrote above should be build to replicate the 'error'.

    cardid,Product
    10150,softdrink
    10150,fruitveg
    10236,frozenmeal
    10236,beer
    10360,fish
    10360,cannedveg
    10360,beer
    10360,frozenmeal
    10451,confectionery
    10451,frozenmeal
    10451,beer
    10451,cannedveg
    10609,fish
    10609,fruitveg
    10614,softdrink
    10645,fruitveg
    10645,frozenmeal
    10645,beer
    10645,cannedveg
    10645,freshmeat
    10717,fish
    10717,fruitveg
    10717,freshmeat
    10872,fish
    10872,frozenmeal
    10872,cannedveg
    10872,beer
    10902,fruitveg
    10902,wine
    10915,fruitveg
    10915,cannedmeat
    10915,fish
    10915,dairy
Sign In or Register to comment.