Pokémon Data Warehouse & Analysis (MongoDB Project)

Master’s Program – NoSQL / MongoDB Module

Author: Carlos Leon Environment: Python 3.12 · Jupyter Notebook · MongoDB (local instance)

---

Overview

This project demonstrates a complete data pipeline and analytical workflow using MongoDB and Python.
It sources live data from the PokéAPI, stores it in MongoDB, performs modeling, transformation, and aggregation, and generates both analytical results and visualizations.

The goal is to build a Pokémon Data Warehouse, similar to a biological or ecological study database — enabling exploration of Pokémon species characteristics, habitats, and relationships.

The notebook for this project can be seen in the pokeapi_to_mongodb.py here, and as a pdf version here

---

Project Workflow

1. Data Collection

• Connected to PokéAPI endpoints:
  • /pokemon-species/ – species-level biological details (habitat, color, growth rate, etc.)
  • /pokemon/ – individual Pokémon attributes (stats, abilities, height, weight, types)
  • /pokemon-habitat/ – contextual grouping of species
• Data fetched via Python using requests and stored raw in MongoDB for cleaning and transformation.

2. Data Modeling

Three main MongoDB collections were created:

Collection	Description
species_clean	Cleaned and standardized Pokémon species data
pokemon_clean	Cleaned Pokémon-specific data (stats, types, abilities)
pokedex_entries	Simplified dataset emulating a real Pokédex (name, type, height, flavor text, sprite)

These are linked logically via common fields like name and dex_number.

3. Data Cleaning & Transformation

• Extracted relevant fields from nested JSON objects.
• Removed duplicates and missing records.
• Created derived attributes (height_m, weight_kg, etc.).
• Enforced consistent naming conventions and standardized field structures.

4. Data Visualization

Key visuals included:

• Pokémon species by habitat and status (regular, legendary, mythical)
• Dual-type co-occurrence analysis
• Average base stats by primary type
• Evolution chain sizes
• Egg group × habitat heatmap
• Interactive Pokédex entry card visual for a selected Pokémon

Each visualization was generated directly from MongoDB aggregation results using matplotlib and pandas.

5. Aggregation Pipelines

1. Average Weight (kg) by Habitat Distribution
2. Habitat distribution by legendary/mythical status
3. Type co-occurrence network
4. Average base stats per type
5. Evolution chain sizes
6. Egg-group × habitat frequency analysis

Each pipeline demonstrates different aggregation operators (e.g., $match, $group, $project, $unwind, $sort, $cond, $arrayToObject).

6. Pokedex Simulation

A lightweight “Pokédex” collection was built to visualize Pokémon entries with:

• Species name and genus
• Type and physical data
• Flavor text
• Sprite image and Artwork

An inline visual cell mimics an actual Pokédex entry.

7. Schema Diagram

A class diagram was created using Mermaid, representing:

• Collections (species_clean, pokemon_clean, pokedex_entries)
• Their logical relationships (name / species_id links)

---

Technologies Used

Category	Tools
Programming	Python 3.12
Database	MongoDB (local instance) + Compass
Visualization	Matplotlib
Development	VS Code + Jupyter Notebook
API	PokéAPI
Class Diagram	Via mermaidchart.com