**The most surprising finding is how little of this dataset's funding "total" is made up of ordinary deals. A handful of mega-deals, including one that looks like a data error, dominate it.** - **Concentration:** Of the 2,065 deals with a usable USD amount, the total is about **$38.06B**, but the **median deal is only $1.7M**. 73 deals of $100M or more (about 3.5% of valued deals) account for **63.6% of all dollars** (step 10 / step 12). The mean works out to roughly $18M, about 10 times the median. Any "average ticket size" or "total funding by vertical, city or year" figure is therefore driven by a few outliers, not by the typical startup. - **One deal is 10% of everything:** The largest record is **Rapido Bike Taxi, Series B, 27/08/2019, $3.9B**. That alone is **10.2% of all dollars** (step 1) and is bigger than Flipkart's biggest deal ($2.5B in 2017). A $3.9B Series B for a bike-taxi startup is implausible. It looks like a unit or entry error, and I could not verify it from the data alone. If it is wrong, the headline total and every rollup that includes it are inflated. - **One bad row breaks simple sums:** The "Drums Food" row (21/07/2016) has an amount that parses as a non-finite number (`nan`). That is why several of my aggregate queries returned `nan` for totals and maxima (steps 0, 2, 4–9, 13), and why the clean runs gave 2,065 valued rows with 1 flagged non-finite row (step 8). A naive `SUM` or `MAX` on this column returns garbage. - **A third of the data has no amount:** 978 of 3,044 rows (about 32%) have no usable amount (step 5 / step 13). Any funding-size analysis therefore covers only about two-thirds of the deals. **Takeaway:** In this dataset the typical deal is small, about $1.7M. The dollar totals are dominated by a few large deals, at least one of which is probably mis-entered. Use medians, or exclude or verify the $100M+ deals, before drawing conclusions about industries, cities or trends from summed dollars.