EDA Practitioner
EDA with Python, Pandas and Jupyter · v30.8
Updated 1 month, 2 weeks ago · Created 1 month, 2 weeks ago
Description
The person or AI system running this playbook end to end — analyst, programmer, and domain expert by turns. Frames the question, does the pandas/stats/ML work, designs the visuals, and builds the Plotly notebook. Reports findings to the Data Sponsor (the external stakeholder who receives the final notebook/report but performs no playbook activity).
Details
- Created:
- Sep 18, 2026
- Updated:
- Sep 18, 2026
Playbook
Used in Activities
- Design and run a predictive experiment Assess Information Gain and Conclude
- Evaluate model quality Assess Information Gain and Conclude
- Interpret correlation with p-value and rule out spurious relationships Assess Information Gain and Conclude
- Rank features by information gain and draw conclusions Assess Information Gain and Conclude
- Use clustering as an alternative when correlation is weak Assess Information Gain and Conclude
- Design a correlation display Choose Chart per Analysis Type
- Design a deviation display Choose Chart per Analysis Type
- Design a distribution display Choose Chart per Analysis Type
- Design a multivariate display Choose Chart per Analysis Type
- Design a part-to-whole/ranking display Choose Chart per Analysis Type
- Design a time-series display Choose Chart per Analysis Type
- Add reference lines and regions Encode and Reduce Clutter
- Check for spurious or missed patterns Encode and Reduce Clutter
- Choose pre-attentive encodings by precision hierarchy Encode and Reduce Clutter
- Identify the analytical pattern type per conclusion Encode and Reduce Clutter
- Plan trellis or multi-view layout Encode and Reduce Clutter
- Reduce over-plotting Encode and Reduce Clutter
- Select the optimal quantitative scale Encode and Reduce Clutter
- Choose candidate target variable(s) Y Explore and Engineer Features
- Compute correlation/covariance and significance between features and Y Explore and Engineer Features
- Compute descriptive statistics per candidate variable Explore and Engineer Features
- Cross-tabulate categorical candidates against Y Explore and Engineer Features
- Engineer and derive features Explore and Engineer Features
- Assess analyst readiness Frame the Problem
- Assess dataset quality against meaningfulness traits Frame the Problem
- Choose the analysis type Frame the Problem
- Formulate the driving questions Frame the Problem
- Select candidate datasets from the datalake catalog Frame the Problem
- Write the Framing Brief Frame the Problem
- Apply scale/reference-line/annotation rules via layout methods Implement Plotly Visuals
- Assemble the Jupyter storybook notebook Implement Plotly Visuals
- Build trellis/small-multiples and multi-view linking Implement Plotly Visuals
- Implement each storyboard entry as a Plotly figure Implement Plotly Visuals
- Review and export the final notebook/report Implement Plotly Visuals
- Set up the Plotly/Jupyter environment Implement Plotly Visuals
- Combine and deduplicate datasets Organize and Clean Data
- Handle missing and invalid data Organize and Clean Data
- Inspect structure Organize and Clean Data
- Load selected datasets into DataFrames Organize and Clean Data
- Reshape data for the framing questions Organize and Clean Data
- Draft one storyboard entry per chart Storyboard the Notebook
- Review the storyboard against conclusions Storyboard the Notebook
- Sequence the narrative Storyboard the Notebook