Three questions stand out. I haven't run the analyses behind them, so I've grounded each in what the dataset can support and where its data quality limits it. ## 1. Is funding activity accelerating or slowing, and which sectors drive the change? - **Why it matters:** this is the timing question for founders raising money and for investors deciding when to enter a market. - **What the data offers:** 3,037 of the 3,044 deals have a parseable date, and the latest is 2020-01-13. That allows deal counts and total dollars by month or quarter, split by industry vertical. - **Caveats:** - The minimum parsed date is `0015-07-01`, which looks like a two-digit-year parsing error. Those dates need cleaning before any trend is trusted. - `Industry Vertical` has 822 distinct values, so it needs grouping into a handful of themes (for example fintech, e-commerce, consumer internet) before sector trends are readable. ## 2. Which cities and verticals attract the most capital, and how concentrated is it? - **Why it matters:** this guides where to launch, open an office, or source deals. - **What the data offers:** 115 distinct cities and 822 verticals. Comparing deal count with total and median ticket size by city, and by city × vertical, would show whether capital is concentrated in a few hubs. It would also show whether some cities have many deals but small cheques. - **Caveats:** - Deal count and dollars can tell different stories, so compare both. - City names are likely inconsistent in spelling, for example "Bangalore" and "Bengaluru". ## 3. What does a typical raise look like by round type and sector, and who funds it? - **Why it matters:** this answers "how much should we raise?" and "which investors should we approach?". - **What the data offers:** 57 distinct funding round types (`InvestmentnType`), and every deal has an `Investors Name` value (3,044 of 3,044). The median and spread of amounts by round type, then by vertical, give benchmarks. Splitting out the investors on each deal shows which ones are most active in a given stage or sector. - **Caveats:** - Only 2,066 of 3,044 amounts (about 68%) are numeric. The other 978 are undisclosed or malformed, so benchmarks reflect disclosed deals only and may skew toward larger or more public rounds. - The 57 round labels need merging (Seed, Seed Round, Angel, and so on). - Investors are stored as a delimited text field per deal, so they must be split before counting. ## Why these three Each maps to a decision the dataset can serve: timing, location and sector, and round size and investor targeting. All three need the same cleanup first: fixing dates, merging category labels, and handling the 32% of deals with no numeric amount. I'd start with question 1, since it needs only the date and a coarse vertical grouping.