results correction

grafikbggrafikbg Member Posts: 14 Contributor I
edited December 2018 in Help

hi there, what is the polite act in a case such as: "well rapidminer you made the "following" prediction, there are "real" results, so please correct yourself next time you nerd"

Tagged:

Answers

  • rfuentealbarfuentealba Moderator, RapidMiner Certified Analyst, Member, University Professor Posts: 568 Unicorn

    Hi @grafikbg,

     

    Given the nature of the words you are using, it looks like you are frustrated by something, but we can't know what is it. It is natural, I feel your pain and sorrow, have been in your place thousands of times and wanted to throw everything from my desk. Please, feel free to ask us any question you have so that we can do our best to help you get out of the problem. However, we should do this respectfully. Calm down, take a deep breath, explain the problem (it helps assuming that we are 5-year old kids who don't know anything about your business and your project), share your XML process and let's see what is going on. Remember that we are willing to help if it is in our hands.

     

    Warm Regards,

  • grafikbggrafikbg Member Posts: 14 Contributor I

    hi Rodrigo, and thank you for your warm replay... well i am not that frustrated, just try to use humour, it's makes the working looks nicer:) i am using deep learning to make a simple prediction of binominal atribute. somehow i believe that if i meticulously corect the alghoritm after a while it will improve the results. i will be very thankfull if you help me in this mater.

  • rfuentealbarfuentealba Moderator, RapidMiner Certified Analyst, Member, University Professor Posts: 568 Unicorn

    Hi @grafikbg,

     

     Sure, do you mind to share your XML process with us, so that we can see what are you doing? Also, we would need a more elaborate explanation on what are you modeling.

     

    All the best,

     

    Rodrigo.

  • grafikbggrafikbg Member Posts: 14 Contributor I

    thank you for your generous offering rodrigo, but sharing the XML means you to do the job and this will be an easy victory... i would like to sweat a little and to do it muself, thus i will become familiar with the logic. i would try to explain it, do not judje me harsh if i sound like 5 years old:) there are attributes numbered by it's secuense in the circuit, than rows which are the examples of each attribute status /on or off/ simple like that, i would like to predict the status of each attribute based on all previous examples, they are somehow interconected and there is a pattern. 

  • rfuentealbarfuentealba Moderator, RapidMiner Certified Analyst, Member, University Professor Posts: 568 Unicorn

    Hi @grafikbg,

     


    @grafikbg wrote:

    thank you for your generous offering rodrigo, but sharing the XML means you to do the job and this will be an easy victory...


    Believe me, I don't want to do your job. But I do want to get you out of your issues. Piece of advice: In life, having knowledge is equally important to having the phone number of the one who has the knowledge ;)

     


    @grafikbg wrote:

    I would try to explain it, do not judje me harsh if i sound like 5 years old :)

     

    Nothing could make me happier than someone who tries to explain like I'm 5 years old. It's much easier to understand the overall situation ;)

     


    @grafikbg wrote:

    There are attributes numbered by it's secuense in the circuit, then rows which are the examples of each attribute status /on or off/ simple like that, i would like to predict the status of each attribute based on all previous examples, they are somehow interconected and there is a pattern. 


    I remember this issue. IIRC, @lionelderkrikor gave you a few ideas.

     

    Well, I searched your profile to review the previous question, and my sensei @kypexin had the same questions I have for your problem:

     

    • Do you have a "complete dependency relation" (every switch depends on the others), a "partial dependency relation" (some switches depend on others), or these all are independent?
      • If you have a complete dependency relation between switches, then you don't have enough data. You have to retrieve more data.
      • If you have a partial dependency relation between switches, then you can select the switches that are related and study these separately. You might still lack enough data in some cases.
      • If you don't have a relation between switches (or even with the previous cases in a next iteration), I would decorate it with more information such as... "this switch controls energy in the floor 16, this switch controls energy on engine 23, 500W, 16amp, this switch controls energy on engine 24, 400W, 32amp... (I'm making this up, probably you can come up with better information). Adding more information about certain pieces will give some more context and help you discover why some switches last a week and some others don't. You might find that switches from trademark A last longer than switches from trademark B in certain conditions, for example.

    Let's go step by step, sometimes it's better to talk with someone else about your problem ;)

     

    All the best,

     

     

  • grafikbggrafikbg Member Posts: 14 Contributor I

    thank you rodrigo, yes this is me with my specific task... when i think it over it seems that i must consider the atributes as independent... regarding "more data" i am afraid that this is what i have... but would you /for sciense sake/ help me put the correct results at the right place and to see waht's will happen. something like /open the process in the operator x put the numbers in the right column..../ i will buy you a big chocolate:)

Sign In or Register to comment.