Design and run a predictive experiment
ACE-2 Order: #2 Explore Data and Find Signal Has Dependencies
Updated 1 month, 2 weeks ago
Guidance
Split the feature table 70:30 train/test (seed the RNG first). Choose the model by Y's type (per the Target Variable (Y) Definition from EFS): continuous Y → OLS LinearRegression, or Ridge if predictors are collinear; categorical/binary Y → LogisticRegression (use cut() first to quantize a continuous proxy into labels if needed, and set C well above the default 1.0, e.g. ~20). Fit on training data only, engineering nonlinear predictors (squares/logs/interaction terms) first if a plain linear fit under-captures curvature, then apply the fitted model to the held-out test set. Never touch the test data while fitting.
Details
- Order:
- #2
- Phase:
- Predecessor:
- ACE-1 Interpret correlation with p-value and rule out spurious relationships
- Successor:
- ACE-3 Evaluate model quality
- Created:
- Sep 18, 2026
- Last Updated:
- Sep 18, 2026
Workflow
Assess Information Gain and Conclude
Rigorously test whether the feature table actually carries information about Y — via significance-tested correlation, a validated predictive experiment, and …
View WorkflowAssigned Agent
EDA Practitioner
The person or AI system running this playbook end to end — analyst, programmer, and …
Required Skills
- Predictive Experiment Design and Model Validation Machine Learning Methodology scikit-learn
- Regression and Classification Modeling Predictive Modeling scikit-learn
Rules
-
Recommended train/test split is 70:30
recommended-traintest-split-is-7030
Input Artifacts 2
-
Feature Table
Document
Required
Produced by: Engineer and derive features -
Target Variable (Y) Definition
Document
Required
Produced by: Choose candidate target variable(s) Y
Output Artifacts
No output artifacts