We look forward to your participation!

Predictive Server Beta

Thanks for participating in the Predictive Server Beta for Alteryx associates! Find resources, ask questions, share ideas, and collaborate with fellow participants.

Not sure where to start? CLICK HERE.
Beta Walkthrough

Missed the live session? No problem! Check out the recorded version for an in-depth overview on how to participate.

Watch Video

Predictive Server Feedback with Kaggle Tabular Data Playground

TimothyL
Alteryx Alumni (Retired)

Beta 1 trial (Jan)

 

Key Selling Point (What Designer cannot do):

  • Better Correlation Charts. 
  • Auto Feature Engineer
  • More Tree-Based Models

 

Key Issues (What Designer can do but PS cannot):

  • Model Interpretation for Regression. Eg. QQ plot. 
  • Hyper Parameters setting are still fixed at the moment.
  • Computation Tunning. 
  • Feature Tools & Predictive Server still not smoothly integrated. (Details below)

 

Details for Data Health: 

  • Outlier Definition 
  • Well Distributed Definition
  • Rows Null discrepancies. The data is without empty rows but it indicates null rows.

 

Detials for Correlation:

  • For Chord Diagram display. It appears with threshold on 0.4 which can be empty page if no strong correlated variables. 

 

Details for Auto Insight:

  • The Input Variable under Target setting is empty. & The General Setting cannot be change
  • The Advanced Setting Under Feature Engineer, takes 2 minute to show the feature tools list.
  • Auto Feature Enginner do appear error due to value condition (believe some transformation / aggregation only accepts certain data structure.

Details for Model Interpretation:

  • Under Model Insights, when switch to another model, the Holdout option is grey out and have to be retrained.  Hard to compare the feature importance between models
  • More peformance charts could be added for Regression

Thanks for the amazing work and design. Looking forward to the next beta. 

TL

0 REPLIES 0