Here are three questions this dataset can answer well, each with the caveat that affects it. The dataset has 3,044 reported deals across 115 cities, 822 verticals, and 57 round types (step-0/1). ## 1. Is Indian startup funding accelerating or slowing, and where did the peak fall? Deal counts by year (step-2) show a clear arc: - 929 deals in 2015 and 993 in 2016 (the peak). - 687 in 2017, 309 in 2018, and 111 in 2019. - Only 7 in 2020, which is probably a partial year because the latest date is 2020-01-13. This looks like a sharp contraction after 2016. It matters for timing a fundraise or market entry, and a natural follow-up is whether total dollars fell as fast as deal counts, or whether fewer but larger rounds replaced many small ones. **Caveats:** - The data may reflect reporting coverage rather than real activity, since it is a compiled deal list. - Only 2,066 of 3,044 deals (about 68%) have a numeric amount (step-0/1). Any dollar-based trend rests on that subset. - Some dates are unusable. One row has the year 0015, and 7 rows did not parse (step-2). ## 2. Which cities and verticals attract the most capital, and what is a typical round for each? The dataset has 115 cities and 57 round types, so you can rank locations and sectors by deal volume and by median ticket size. You can also benchmark typical raises, for example seed versus Series A for a given vertical in a given city. This serves founders deciding where to launch or how much to raise, and investors deciding where to source deals. **Caveats:** - Round types and vertical labels are free text, so they need normalising first. For example, variants of "Seed" or "Private Equity" will split counts. - Medians are better than means here, because a few very large deals will skew averages. ## 3. How concentrated is the investor network, and who backs which sectors? Every one of the 3,044 deals has an investor field (step-0), and 2,459 distinct startups appear. Many startups therefore raise more than once. Two things could be built from this: - **Repeat investors:** splitting multi-investor strings would show which investors are most active, and in which sectors, cities, and stages. - **Follow-on funding:** tracing startups across their rounds would show how often early-stage companies reach later rounds, and how long that takes. This would help founders decide whom to approach, and ecosystem analysts see how dependent the market is on a few players. **Caveats:** - Investor names are inconsistently spelled and combined in a single column, so they need cleaning and splitting first. ## A data-quality warning that affects all three The vertical field changes character over time. There are 724 distinct verticals in 2015 but only 19–53 per year from 2016 onward (step-2). That suggests the 2015 data uses free-text labels while later years use a coarser or standardised taxonomy. Sector-over-time comparisons across 2015 and 2016 should be made only after mapping labels to a common scheme. Otherwise a "shift" in sectors may just be a change in how deals were labelled.