Please enable JavaScript.
Coggle requires JavaScript to display documents.
Automating Machine Learning (8 criteria for AutoML Excellence…
Automating Machine Learning
Machine Learning Life cycle
Model Data
Interpret & Communicate
Aquire & Explore Data
Implement, Document & Maintain
Define Project Objectives
Machine learning is about automating complex tasks that require speed and accuracy
AutoML: important areas
Feature engineering
Algorithm Selection and hyperparameter turning
Exploratory data analysis
Model Diagnostics
Available Tools and Platforms
Context-specific tools
General platforms
Open-source
Commercial
8 criteria for AutoML Excellence
Productivity
Ease of Use
Accuracy MOST IMPORTANT
Understanding & Learning
Resource Availability
Process transparency
Generalized across contexts
Recommend actions
CH3 Specify Business Problem
Problem statements: evaluate off of 3 criteria
CH4 Acquire Subject Matter Expertise
Important to identify potential obstacle or opportunities
Decide on Unit of Analysis
Who
Where
What
When
Determining the Unit of analysis
Figure out what the prediction target is
Ch6 Define Prediction Target
Prediction target is the behavior of a "thing"
The importance of Target in ML
Without a target there is no way we can learn what associations are made within data
Success, Risk and Continuation
Identify Success Criteria
Is mgmt on board with the project
Can the model Drivers be Visualized
Who will use the model
how much value can the model produce
Foresee Risks
Ethical Risks
Cultural risks
Model Risks
environmental risks
Decide Whether to Continue