How lalitnayyar/iimkmodule4assignment works
Main components
a. Data Sources
• Online Retail.xlsx (raw transactional data)
• cleaned_retail_data.csv (intermediate cleaned data)
b. Data Processing & Cleaning Layer
• LalitNayyarIIMKMod4_analysis_fin.ipynb (data profiling, cleaning functions)
• fixnotebookscripts/ (a library of Python scripts to auto-fix notebooks)
c. Analytics Notebooks (core analysis stages)
• Descriptive: LalitNayyarIIMKMod4_descriptive_analysis_fin.ipynb
• Diagnostic: LalitNayyarIIMKMod4_behavior_diagnostic_analysis_fin.ipynb
• Predictive: LalitNayyarIIMKMod4_predictive_analysis_fin.ipynb
• (plus transitionnotebooks/ historical versions)
d. Submission Generation Layer
• submissionpython/ (scripts that convert notebooks to HTML, apply templates & styling)
e. Environment & Configuration
• requirements.txt (Python dependencies)
• Jupyter execution environment (runtime)
Relationships & data flow
1. Raw data (Excel) → Data preprocessing notebook
2. Preprocessing → cleaned_retail_data.csv
3. cleaned_retail_data.csv → Descriptive, Diagnostic, Predictive notebooks
4. Each analysis notebook outputs figures, tables, and intermediate results
5. fixnotebookscripts/ can be invoked at any notebook stage to standardize formatting or fix metadata
6. submissionpython/ consumes final notebooks → produces HTML submissions (with themes, TOC, contrast, etc.)
Architectural patterns & design principles
• Pipeline architecture: sequential flow (Ingest → Clean → Analyze → Publish)
• Modularization: separation of concerns
• Cleaning code in one notebook + reusable Python scripts
• Analysis split by maturity stage
• Presentation code in dedicated scripts
• Extensibility: drop-in additional analysis notebooks or submission templates without disrupting core pipeline
• Reproducibility: versioned notebooks + requirements.txt + scripted conversion
Relevant technologies/frameworks
• Python ecosystem: pandas, numpy, matplotlib, seaborn, scikit-learn
• Jupyter Notebook / nbconvert
• Custom notebook fixer scripts (pure Python)
• HTML/CSS templates (via submissionpython scripts)