Hitachi Vantara Pentaho Community Forums
Results 1 to 3 of 3

Thread: Time series forecasting in KnowledgeFlow

  1. #1
    Join Date
    Jan 2016
    Posts
    2

    Default Time series forecasting in KnowledgeFlow

    Hello everyone,



    I’m working on a project for my university studies, I am trying forecast the future revenue of this dataset:

    @relation rest-forecasting
    @attribute day DATE 'yyyy-MM-dd'
    @attribute days { Mon, Tue, Wed, Thu, Fri, Sat, Sun}
    @attribute ricorrenza { yes, no}
    @attribute event { dj, live, djlive, none}
    @attribute cost numeric
    @attribute revenue numeric
    @data
    2014-06-01,Sun,no,dj,140,1913
    2014-06-02,Mon,yes,none,80,807.5
    2014-06-03,Tue,no,none,0,505.5
    ………




    Because I have overlay data, I read in this forum thati will work in KnowledgeFlow environment (or PDI) and import the datasetwith training/priming data followed by data to be forecasted (set as missing value).
    I have never worked with knowledgeFlow, I don’t know how to produce some reports in this environment with TimeSeriesForecasting. Particularly, I am not be able to connect TimeSeriesForecasting with TextViewer (because there is no possibility).


    If anyone could be of some help I would truly appreciate it.
    Thanks,
    Danilo

  2. #2
    Join Date
    Aug 2006
    Posts
    1,741

    Default

    The TimeSeriesForecasting step in the KnowledgeFlow imports a time series model (created in the time series environment) and uses it to produce predictions for incoming data. It produces a stream of data as output (historical priming instances/rows followed by instances containing forecasted values) via an "instance" connection in the Knowledge Flow. This can be connected to a data sink in order to save the data to a file (e.g. CSV or ARFF file). Alternatively, you could connect it to an InstanceStreamToBatchMaker step, which in turn can be connected to a TextViewer via a "dataSet" connection.

    Cheers,
    Mark.

  3. #3
    Join Date
    Jan 2016
    Posts
    2

    Default

    Quote Originally Posted by Mark View Post
    The TimeSeriesForecasting step in the KnowledgeFlow imports a time series model (created in the time series environment) and uses it to produce predictions for incoming data. It produces a stream of data as output (historical priming instances/rows followed by instances containing forecasted values) via an "instance" connection in the Knowledge Flow. This can be connected to a data sink in order to save the data to a file (e.g. CSV or ARFF file). Alternatively, you could connect it to an InstanceStreamToBatchMaker step, which in turn can be connected to a TextViewer via a "dataSet" connection.

    Cheers,
    Mark.
    Now i have the result! Thanks a lot Mark.

    Danilo

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  
Privacy Policy | Legal Notices | Safe Harbor Privacy Policy

Copyright © 2005 - 2019 Hitachi Vantara Corporation. All Rights Reserved.