Thanks for this video - very helpful overview of the modeling and scoring process in MS. A couple of comments / questions: I have a bit of experience with IBM SPSS Modeler, and I'm curious to understand a bit more about how MicroStrategy works "under the hood" with things like partitioning the data into test, training and validation sets, among other things, so that I can make an informed feature comparison between the two tools. What resources (documentation, video or otherwise) could you recommend in this regard? Regarding the chart in the output, was that a gains chart? Are there options to show lift, ROI, and other metrics around model performance? If you are going to consider doing another video on this, I would be intrigued to see how the decision tree functionality works for categorical and continuous variables, as this is usually one of the more "business-friendly" ways of presenting a model output when trying to "sell" the model to the business users. And...one last question (and then I'll go away, I promise): is there any way to use these models for batch scoring metrics and writing the predicted values to a database for further analysis, or would that have to be done using another tool like SAS? Thanks again for your efforts.
Very loaded comment, but very good topic. Under the hood now you can run R, MicroStrategy allows you to integrate R and use native R scrippting for and mining task. However, this remains a limited feature of MSTR since it has limited scoring capabilities - limited to the metrics rendered by a report. I will look into adding another video for something that shows more categorical scoring. And in regards to your batching you can use the report to create a DataMart from the Predictive Report that is written back to the database, you will need DataMart to database write privileges from the Admin console if your not doing this as an Admin.
Wow ok...how come my prediction changes when I change the year. If this is a 2008 prediction based on 2008 data...then shouldnt the prediction be for 2009 when changing to 2009 data. Or am I missing something here
Awesome explanation.Thank you for your Effort. Could you please upload some videos on Dashboard and Documents.
This video is really helpful. Can you upload another video with real time data so as to understand more specifically.
Hello MicroRooster can u do videos on All Functions like Standard Functions,PLUG-IN PACKAGE FUNCTIONS,operators?
Thanks for this video - very helpful overview of the modeling and scoring process in MS. A couple of comments / questions:
I have a bit of experience with IBM SPSS Modeler, and I'm curious to understand a bit more about how MicroStrategy works "under the hood" with things like partitioning the data into test, training and validation sets, among other things, so that I can make an informed feature comparison between the two tools. What resources (documentation, video or otherwise) could you recommend in this regard?
Regarding the chart in the output, was that a gains chart? Are there options to show lift, ROI, and other metrics around model performance?
If you are going to consider doing another video on this, I would be intrigued to see how the decision tree functionality works for categorical and continuous variables, as this is usually one of the more "business-friendly" ways of presenting a model output when trying to "sell" the model to the business users.
And...one last question (and then I'll go away, I promise): is there any way to use these models for batch scoring metrics and writing the predicted values to a database for further analysis, or would that have to be done using another tool like SAS?
Thanks again for your efforts.
Very loaded comment, but very good topic. Under the hood now you can run R, MicroStrategy allows you to integrate R and use native R scrippting for and mining task. However, this remains a limited feature of MSTR since it has limited scoring capabilities - limited to the metrics rendered by a report. I will look into adding another video for something that shows more categorical scoring. And in regards to your batching you can use the report to create a DataMart from the Predictive Report that is written back to the database, you will need DataMart to database write privileges from the Admin console if your not doing this as an Admin.
Wow ok...how come my prediction changes when I change the year.
If this is a 2008 prediction based on 2008 data...then shouldnt the prediction be for 2009 when changing to 2009 data. Or am I missing something here