## Missing values by column **No column has true NULLs.** The null-count check returned 0 rows, and a separate per-column NULL check showed 0.0% for every column in every month. The missing data is stored as empty or placeholder values instead. For example, a stage value of `[]` appeared 43 times in one stage breakdown. The data is split into five monthly tables, January to May, with 446 rows in total. Counting those empty or placeholder values as missing, here is the pooled result across all five months: | Column | Missing rows | % missing (of 446) | |---|---|---| | stage | 127 | 28.5% | | investor | 45 | 10.1% | | founders | 38 | 8.5% | | amount | 25 | 5.6% | | location | 20 | 4.5% | | company_name | 19 | 4.3% | | sector | 19 | 4.3% | | founded | 18 | 4.0% | | what_it_does | 18 | 4.0% | ## Where the gaps are - **Stage** is the biggest gap, but it varies a lot by month: - January: 37.4% - February: 30.2% - March: 31.6% - April: 3.2% - May: 50.0% - **May is the problem table.** Of its 42 rows, 42.9–50% are missing in every column except founders, which is 64.3% missing. The company, founded, location, sector and what_it_does gaps (18–19 rows each) account for nearly all of the pooled counts for those columns. April has just 1 missing location, and January to March have none. - **Investor** is missing in every month, from 3.1% (March) to 10.5% (April), plus 50% in May. - **Founders** is missing mostly in April (10.5%) and May (64.3%), with 1 missing row in February. - **Amount** is missing in 2 rows each in February and April, and in 21 rows in May. ## What this leaves open - I don't know the exact rule that classified a value as "missing". I inferred it was empty or placeholder values such as `[]`, but I did not confirm the rule for every column. - The stage breakdown I saw (43 `[]` values) matches January's count, so it appears to cover only one month. I can't say what other placeholders exist in the other tables. - I can't say whether the May gaps are genuinely blank records, a partial load, or a different file format. The pattern, with roughly the same 18–19 rows missing across the core fields, suggests a data-quality problem in that month's table.