Only two columns can be treated as numeric: **Founded**, a year, and **Amount**, which is stored as text like "$1,000,000" and was converted to a USD number for the summary. The other columns are text or system metadata. The summary below comes from step-1, which covers all 446 rows across the five monthly tables. | Metric | Founded (year) | Amount (USD) | |---|---|---| | Valid values (of 446 rows) | 427 | 385 | | Min | 1871 | 11.3 | | 25th percentile | 2014 | 1,300,000 | | Median | 2017 | 9,400,000 | | Mean | 2015.3 | 145,567,580 | | 75th percentile | 2020 | 45,000,000 | | Max | 2022 | 18,300,000,000 | | Std deviation | 11.0 | 991,545,057 | **Founded** - The distribution is fairly tight and skews old: the median (2017) is slightly above the mean (2015.3). - The middle half of companies were founded between 2014 and 2020, a span of 6 years. - The standard deviation of 11 years is inflated by a few very old companies, such as the 1871 minimum. - 19 rows have no usable year. The May table stores Founded as text, and the card notes some May rows are misaligned or blank. **Amount (USD)** - The distribution is extremely right-skewed. The mean (about $145.6M) is roughly 15 times the median (about $9.4M). - The standard deviation (about $992M) is nearly 7 times the mean, so a handful of very large rounds dominate it. The maximum is $18.3B. - The median and the 25th–75th percentile range ($1.3M to $45M) describe a typical round far better than the mean does. - Treat the extremes with caution. The $11.3 minimum and the $18.3B maximum look like data-entry or parsing artefacts, for example a non-USD or mis-scaled value, and I did not check them row by row. - 61 rows have no numeric amount. Step-0 shows "Undisclosed" accounts for 6 of the most common values, and the rest are probably blank or misaligned rows. - The most common disclosed amounts are round figures: $1M (9 rows), $10M (8) and $5M (6).