How eppursimuove9/e-commerce-sales-analysis works
Project Type and Purpose
• Type: End-to-end data analysis & visualization pipeline hosted in Google Colab.
• Purpose: Ingest UK retailer transaction data, clean & preprocess it, run EDA & K-Means clustering, then generate interactive dashboards (Plotly) and static visual assets.
File Structure
• assets/: contains chart PNGs.
• ecommerce‐analysis.ipynb: Jupyter/Colab notebook with all code stages.
• requirements.txt: Python dependencies.
No backend server or frontend application—everything lives in the notebook environment.
README for Additional Insights
• Data source is an Excel file auto-downloaded from Google Drive.
• Tech stack: Python, Pandas/NumPy/SciPy, Scikit-learn, Plotly/Matplotlib/Seaborn.
• Workflow: Data cleaning → EDA → ML clustering → Dashboard export.
Main Components and Their Interactions
a. Data Source
– Online-Retail.xlsx stored on Google Drive.
– External: Plotly Dash hosting (for live dashboard link).
b. Compute Environment
– Google Colab VM running Python.
– Dependencies installed via requirements.txt.
c. Notebook Modules (sub-components inside Colab)
1. Data Ingestion & Download Script
2. Data Cleaning & Preprocessing (Pandas)
3. Exploratory Data Analysis (Matplotlib, Seaborn)
4. Machine Learning (Scikit-learn K-Means, Elbow Method)
5. Dashboard Generation (Plotly Python API)
6. Asset Export (PNG & HTML files)
d. Outputs & Hosting
– Interactive dashboards published to Plotly’s cloud.
– PNG visual assets in /assets.
– Standalone HTML files downloadable from Colab.
Architectural Patterns and Design Principles
• Modular notebook cells as functional stages.
• Separation of concerns: data logic vs. modeling vs. visualization.
• Externalize long-term storage (Google Drive) and dashboard hosting (Plotly).