UT Dallas MSBA capstone (Group 7) - A strategic analytics case study on Conagra's Gardein brand in the U.S. frozen meat-substitute category. Benchmarks Gardein vs Morningstar Farms and the broader competitive set using multi-year retail POS data, multivariate regression on dollar sales, and geographic opportunity mapping.
📄 Deliverables: 🔗 Final Report (PDF) · 🔗 Executive Deck (PPTX)
Gardein sits in a crowded and rapidly shifting plant-based frozen category. Management needed to understand:
- Where is Gardein winning and losing relative to Morningstar Farms and the broader competitive set?
- Which product and pricing levers (pack size, price per unit, distribution, meat source, flavor) most influence dollar sales?
- Which U.S. geographies offer the strongest growth opportunity - and why?
| Stage | Methods |
|---|---|
| Data | Multi-year retail POS extracts for Gardein, Morningstar Farms, and competitor brands (Nielsen-style syndicated data format) |
| Cleaning | Missing-value treatment, duplicate checks, column pruning |
| Feature engineering | Parsed product description strings → franchise, meat source, product type, flavor, pack size (ounces) |
| EDA | Univariate + bivariate distributions; category and sub-segment share |
| Multicollinearity | VIF review, consolidation / removal of collinear predictors |
| Modeling | Multivariate OLS regression on dollar sales with controls for distribution (ACV), price, pack size, product, brand, meat source, and geography |
| Geographic analysis | Regional cuts (California, Northeast, Total US Multi-Outlet + Convenience) aligned to regression region effects |
📊 Key Findings (see the Final Report for details)
High-level takeaways - for full coefficients, significance tests, and recommendations, see the PDF report and slide deck.
- Dollar sales are driven most strongly by ACV-weighted distribution and price per unit, with meaningful secondary effects from pack size and product type.
- Gardein vs Morningstar Farms positioning diverges on health messaging, ingredient innovation, and sustainability - each with distinct implications for shelf strategy.
- Regional performance shows asymmetric opportunity: certain geographies over-index on the category but under-index on Gardein, flagging clear white-space for distribution expansion.
Export 2–3 key charts (market share, regression coefficient plot, geographic map) from the notebooks or slide deck into a
docs/images/folder and reference them here. Placeholders removed until real exports are added so the README doesn't show broken image icons.
Ashish Patil · Isha Narkhede · Javed Mohammed · Neha Govekar · Pallavi Sawant · Raj Nathwani
[Fill this section in - recruiters specifically look for it on team projects. A few concrete sentences on what you personally owned is far stronger than silence. Examples of angles to cover:]
- What workstream did you lead or co-lead? (e.g., "Led the regression modeling and feature engineering from product description strings")
- What technical pieces are yours? (e.g., "Built the VIF diagnostics and final OLS specification in
reg.ipynb") - What did you present? (e.g., "Authored the geography slides in the final deck and the corresponding section of the written report")
| File | What it is |
|---|---|
FINAL REPORT.pdf |
Written report - objectives, EDA, regression results, recommendations (main deliverable) |
Group_7.pptx |
Executive slide deck - summarized findings for presentation |
reg.ipynb |
Regression modeling notebook - feature engineering, VIF, OLS fit, diagnostics |
Geography_based_visualization.ipynb |
Regional analysis notebook - geography cuts, multi-year panels, maps |
2023_MorningStar_Gardein.xlsx |
Competitor POS data - Morningstar Farms + Gardein |
2023_Other_Brand_Data.xlsx |
Broader category POS data - other competing brands |
# Create a virtual environment
python -m venv .venv
source .venv/bin/activate # macOS / Linux
# .venv\Scripts\activate # Windows
# Install dependencies
pip install pandas numpy matplotlib seaborn jupyter openpyxl scipy statsmodels
# Open the notebooks
jupyter notebookThe notebooks expect the Excel files in the same directory. If your copies live elsewhere, update the pd.read_excel(...) paths at the top of each notebook.
The POS extracts included here are syndicated retail data shared for coursework. Before forking or redistributing:
- Confirm redistribution is permitted under the original data license
- Consider replacing with aggregated / synthetic samples for public demo purposes
- Clear notebook outputs (
Cell → All Output → Clear) if any contain internal identifiers
This repository is an academic / course deliverable. It is not an official Conagra or Gardein publication and the findings should not be interpreted as company positions.
Isha Narkhede · Portfolio · LinkedIn · ishajayant207@gmail.com