🏬 Retail Store Risk Analysis

Predicting Walmart & Target store closure risks using ML + NLP on 1.5M+ Yelp reviews.

---

Problem

U.S. retailers face costly last-minute closures due to lagging indicators (revenue decline, lease expiry, low traffic).
Goal: Predict at-risk Walmart & Target stores early using customer review data.

---

Method

• Data: Yelp reviews (1.5M rows, 275 stores across U.S.)
• EDA: Review length, unique words, sentiment distribution
• Models: Logistic Regression, Naïve Bayes, SVM, Random Forest (hypertuned)
• NLP: Sentiment analysis, BERT topic modeling, zero-shot validation
• Business Analysis: ROI/NPV estimation for proactive interventions

---

Results

• Random Forest (Target): 96.7% accuracy
• Random Forest (Walmart): 94.0% accuracy
• Sentiment split: Target (77% positive), Walmart (60% negative)
• BERT topics: inventory issues, checkout delays, staff quality, cleanliness, security
• Financial impact: $1.25B potential savings; ROI > 1400%:contentReference[oaicite:6]{index=6}

---

Business Insights

• Walmart: Higher closure risk, negative sentiment on service, stock, and online orders.
• Target: Stronger brand sentiment, but pain points in parking, cleanliness, and service.
• Recommendations: Improve service training, inventory management, cleanliness, and security.

---

Future Work

• Extend analysis to multi-retailer datasets
• Incorporate external data (leases, traffic, economic indicators)
• Deploy early warning dashboards for proactive monitoring