Technology: Advanced Analytics

Ranking

• Bayesian weighted rating with global mean C and threshold m
• Popularity via log1p(review_count)
• rank_score = bayes_score * (1 + 0.15 * popularity)

Investor Insights Analytics

• Market Opportunity: filter categories with avg_rating ≥ 4.0 and business_count < 5; score = (avg_rating * avg_review_count) / (business_count + 1)
• Location Hotspots: KMeans clustering on lat/lng; per-cluster stats (dominant city, avg rating, density)
• Competitor Benchmark: avg/median rating, review volume, count, most common price tier

Data Cleaning Highlights

• Coerce numeric columns; fill missing strings; ensure required columns exist
• Transform categories into list for filtering/exploding

Visualizations

• Plotly charts (histograms, bars)
• PyDeck map with color-coded clusters and tooltips