**Yes, the data has a geographic dimension (`City Location`), and it is heavily concentrated in a few metros.** I merged spelling variants of the same city (Bangalore/Bengaluru, Gurgaon/Gurugram, Delhi/New Delhi) so each city is counted once. The rendered table and chart show the resulting split of the 3,044 deals. - **Bengaluru is the largest hub:** 841 deals, 27.6% of the total. The raw spellings were Bangalore (700) and Bengaluru (141). - **Mumbai is second:** 567 deals, 18.6%. - **Delhi / New Delhi is third:** 455 deals, 14.9%. - **Gurugram is fourth:** 337 deals, 11.1%, so the Delhi NCR cluster is large. Noida adds another 92 (3.0%), which puts the whole NCR cluster at about 29% of all deals. - **Together, Bengaluru, Mumbai, Delhi and Gurugram hold about 72% of deals.** - **A second tier of cities each holds 3–3.5%:** Pune (105), Hyderabad (99), Chennai (97) and Noida (92). - **The long tail is thin:** Ahmedabad (38), Jaipur (30), Kolkata (21), Indore (13), Chandigarh (11), Goa (10), Vadodara (10), Coimbatore (5) and Kanpur (4), each at 1.2% or less. - **Overseas locations are rare:** Singapore has 8 deals (0.3%), and "Pune / US" has 4 in the raw spelling check. **Data-quality caveats** - **Missing location:** 171 rows (5.6%) have the literal value `nan` and another 6 are blank. The raw spelling check also showed 3 rows as `N/A`. So about 6% of deals have no usable location. - **Dirty spellings not fully merged:** some values carry a stray non-breaking-space character, such as `\xa0Gurgaon` (4 rows) and `\xa0New Delhi` (3 rows). My merge did not fold these in, so those cities are very slightly undercounted. - **Multi-city entries:** a few values such as "Pune / US" combine locations rather than naming one city. The queries behind these figures are the city-frequency counts (raw spellings, then merged variants) run on `raw.startup_funding`. {"action":"final","content":"","charts":[{"kind":"hbar","x":"city","series":["pct_of_deals"],"title":"Share of funding deals by city (variants merged)"}]}