Poetry Preferences (SQL)

A small exploratory data project based on my past and present poetry preferences.

poetry project

Why This Project

I chanced upon Love’s Witness: Five Centuries of Love Poetry by Women (edited by Jill Hollis), in 2025.

The depth, range, and nuance of the poems was impressive.

I also noticed it was very different from my earlier poetry preferences, which were predominantly written by a small number of male writers.

It was fascinating to explore some personal favorites as a BI-adjacent project.

Dataset

I created the dataset in Google Sheets.

Here’s my original version.

Includes details like:

  • Poet Name
  • Poem Title
  • Country of Poets

Findings Summary

Using two datasets:

  • recent phase poems
  • earlier phase works

I explored thematic patterns, stylistic tendencies, and country‑level distribution across 29 poets.

The project demonstrates SQL fundamentals including joins, grouping, aggregation, and cross‑table analysis.

Key steps included:

  • Cleaning and standardizing column names
  • Importing CSVs into SQLite Online
  • Running relational joins between poems and poets
  • Comparing themes across phases
  • Analyzing stylistic patterns

This project helped me build confidence with SQL joins and grouping logic, and gave me hands‑on experience troubleshooting real‑world data inconsistencies.


Sample Queries

Query 1: “How many poems did I like because of the poet’s style?”

This shows application of the COUNT() function to return the rows that match the criterion, according to phase.

Query 2: “Which poets dominate each phase?”

This shows grouping + joining + phase comparison.

Note: Edgar Allan Poe would have taken the top spot easily, if I had added more examples for him.

Query 3: “All recent poems with poet info”

This shows usage of a SQL Join to combine data from two tables based on a related column (poet_id).


Top 3 Insights

  1. Distinct emotional phases emerged. Earlier‑phase works leaned heavily toward Zen, longing, and existential themes, while recent‑phase poems showed more relational nuance and ethical reflection.
  2. Stylistic diversity increased over time. Recent‑phase poems showed a wider range of styles (personal, narrative, modernist), whereas earlier‑phase works clustered around Zen, Romantic, and Gothic traditions.
  3. Style preference in phases. The results show that I liked more than 50% of recent poems because of the poet’s style, compared to 20% of earlier poems for the same reason.

Additional Insight: Why Normalization Matters

While working with the resonance_tags column, I realized that many rows contained multiple values inside a single cell (e.g., “style, lyricism, idealism”).

When I tried to group or count individual tags, SQL treated the entire string as one category.

This experience helped me understand why normalization is important: storing one value per cell makes querying, grouping, and analysis much cleaner and more flexible.

For this project, I used LIKE filters to answer my questions, but in a future iteration I may create a separate resonance_tags table (one row per tag) for more streamlined analysis.