Here are the three questions I think this dataset can answer best. They map to the decisions in the dataset's context: where to focus, who funds, and whether activity is speeding up. **1. Is funding activity accelerating or slowing, and which verticals are driving it?** - The data has 3,044 deals, and 3,041 of them have a parseable date. That is enough to build deal counts and total dollars by month or quarter, split by industry vertical. - It would show whether the 2015–2020 period was a boom or a plateau, and whether growth is broad or concentrated in a few sectors. - The profile shows 822 distinct verticals, so verticals will need grouping into major buckets first (for example consumer internet, e-commerce, fintech). - Date quality needs care. The minimum date reads as `0015-07-01`, which looks like a year typo for 2015. The valid range runs to `2020-01-13`, so a few dates must be cleaned before trending. **2. Where does the money go? How concentrated is funding by city and by vertical, and are ticket sizes different in the top hubs?** - The data covers 115 distinct cities. I would compare deal count, total and median amount, and share of the total for each city, then cross that with vertical. - This would show whether funding is dominated by a few hubs and whether other cities specialise in particular sectors. - It serves the "which city to launch in or source deals from" decision. - Use the median rather than the mean, because a few very large rounds will skew averages. **3. Who are the repeat investors, and what round types do they back by sector?** - Every one of the 3,044 deals has an investor listed, and the data has 57 distinct round types across 2,459 startups. - I would split the multi-investor strings into individual investors. Then I would rank them by deal count and see which round types they favour, such as angel, seed, Series A or private equity. - I would also look at which startups raised repeatedly (3,044 deals over 2,459 startups means some did) and how long they took between rounds. Together these show typical funding paths and who to approach at each stage. **Data caveats** - Only 2,066 of 3,044 deals (about 68%) have a parseable dollar amount. Any question about amounts or ticket sizes covers about two-thirds of the deals, and the missing third may not be random. For example, undisclosed amounts could be more common in certain round types. - The vertical and round-type columns are messy, with 822 and 57 distinct values. Spelling variants will need normalising before the breakdowns are reliable. I can run any of these next. I would start with question 1 using a normalised month and vertical grouping.