Here are the three questions I think this dataset can answer best. Each one fits the decisions it could support (where to invest, where to launch, when to raise), and each comes with a caveat about the data. ## 1. Is Indian startup funding accelerating or slowing, and is it moving from many small deals to fewer large ones? This is the timing question for a raise or a market entry. The yearly results point to a clear shift: | Year | Deals | Median disclosed ticket | |---|---|---| | 2015 | ~935 | ~$1.5M | | 2016 | 993 | ~$1.0M | | 2017 | 687 | ~$2.25M | | 2018 | ~310 | ~$3.9–4.0M | | 2019 | 111 | ~$12M | | 2020 | 7 | ~$9M | Deal count falls sharply after 2016, while the median disclosed ticket rises about 8x from 2016 to 2019. Disclosed totals stay in the same range (roughly $5–10B per year), so fewer, bigger rounds carry the money. Caveats: - Only 2,066 of 3,044 deals have a numeric amount. - The 2016 total came back `nan` because a non-numeric value got into the sum, so I can't quote a reliable 2016 total. - The falling deal count may partly reflect reporting coverage rather than real market activity. - 2020 holds only 7 deals, and 2019 holds 111. ## 2. Which cities and verticals dominate deal flow, and where is the money concentrated? This helps with choosing a city to launch in, open an office in, or source deals from. Bangalore (700 deals), Mumbai (567), New Delhi (421) and Gurgaon (287) lead on deal count. Median tickets differ by city: - Kolkata has a median of about $157K. - Bangalore has a median of about $2M. - "Bengaluru" has a median of about $5M. Industry verticals are very fragmented, with 822 distinct values, so a vertical-level answer needs grouping first. Caveats: - City names are not standardized. Bangalore and Bengaluru, Gurgaon and Gurugram, and Delhi and New Delhi are split. Merging them would change the rankings. - 171 deals have no city. - The 822 vertical labels would need to be mapped to a manageable set of categories. ## 3. Which investors and round types are typical for each sector and stage? This helps founders benchmark a raise and decide which investors to approach. Every one of the 3,044 deals has an investor field, and there are 57 distinct round types and 2,459 distinct startups. That is enough to ask: - What does a typical seed, Series A or private-equity ticket look like by vertical and by year? - Which investors repeatedly back which sectors or cities? - Which startups raise multiple rounds, and how fast do they move from one to the next? Caveats: - The investor field is free text with multiple investors per deal, so it needs splitting and cleaning. - The 57 round types include spelling variants. - Tickets are disclosed in only about two-thirds of deals, so benchmarks are biased toward deals that disclosed. ## Data quality to fix first - Dates are stored as text: 3,037 of 3,044 parse, and the rest are malformed. - One date parses as year 15, which is a typo for 2015. - Amounts are text and only 2,066 are numeric. - City and vertical labels are inconsistent. Cleaning these would make all three analyses much more reliable.