fips county geology counties

This is the FIPS code identifying U.S. counties (or equivalent geographies), with all 3235 values unique and no nulls. Values span 1001 to 78030, consistent with state-prefixed county codes, and the distribution is broad (IQR 27090) rather than meaningfully skewed (skew 0.17). Treat the numeric stats as incidental — magnitude has no quantitative meaning here. Treatment: Cast to string and use as a join key to county-level reference data; do not model as numeric. high · anthropic:claude-opus-4-7

n

3,235

nulls

0 (0.0%)

unique

3,235

min

1,001

max

78,030

mean

3.152e+04

median

30,035

std

1.643e+04

q1

19,036

q3

46,126

iqr

27,090

skew

0.1738

kurtosis

-0.6075

n_outliers

0

outlier_rate

0

zero_rate

0

This column holds US county-level place names, with 1,973 unique values across 3,235 rows and almost every entry containing the word 'county' (2,999 occurrences) alongside Louisiana 'parish' (64) and Puerto Rico 'municipio' (78) variants. Names repeat heavily — duplicate rate is 39% with classics like 'Washington County' (30), 'Jefferson County' (25), and 'Franklin County' (24) topping the list, which is expected since the same county name recurs across states. Entries are short (mean 14.2 chars, ~2 words) and there are no nulls or empties. Treatment: Pair with a state column to form a unique geographic key before joining or aggregating. high · anthropic:claude-opus-4-7

n

3,235

nulls

0 (0.0%)

unique

1,973

len_min

4

len_max

46

len_mean

14.18

len_median

14

len_p95

18

word_mean

2.084

word_median

2

n_empty

0

n_duplicates

1,262

duplicate_rate

0.3901

vocab_size

1,973

readability_flesch_mean

33.65

emoji_rate

0

url_rate

0

one_word_rate

0.0003091

allcaps_rate

0

boilerplate_rate

0

This is a US state code column with 56 unique values — more than the 50 states, suggesting territories or codes like DC, PR, or military designations are included. The distribution is fairly even (entropy ratio 0.92), with TX leading at 7.9% (254 of 3235 rows) followed by GA, VA, KY, and MO, consistent with a county- or jurisdiction-level dataset where larger states contribute more rows. No nulls. Treatment: One-hot or target-encode for modelling; verify the 6 extra codes beyond 50 states. high · anthropic:claude-opus-4-7

n

3,235

nulls

0 (0.0%)

unique

56

top_value

TX

top_rate

0.07852

cardinality

56

entropy

5.338

entropy_ratio

0.9192

This column holds U.S. state names, almost certainly one row per county or county-equivalent given the 3,235 total rows and 56 distinct values (the 50 states plus territories/DC). Texas dominates at 254 rows (7.85%), followed by Georgia (159) and Virginia (133), which matches the known county-count ranking. Distribution is highly even across categories (entropy ratio 0.92) with no nulls. Treatment: Use as a categorical grouping key; one-hot or target-encode for modelling. high · anthropic:claude-opus-4-7

n

3,235

nulls

0 (0.0%)

unique

56

top_value

Texas

top_rate

0.07852

cardinality

56

entropy

5.338

entropy_ratio

0.9192

Numeric feature measuring distance to a deposit, likely in metres, with all 3235 rows populated and 2202 distinct values. The distribution is severely right-skewed (skew 7.51, kurtosis 77.6): the median is 152.0 while the mean is 230.12 and the max stretches to 5652.4, more than 14x the Q3 of 235.75. About 4.9% of rows (159) flag as outliers, and there are no zeros or nulls. Treatment: log-transform before modelling to tame the heavy right tail. high · anthropic:claude-opus-4-7

n

3,235

nulls

0 (0.0%)

unique

2,202

min

1.8

max

5652

mean

230.1

median

152

std

399.9

q1

85.5

q3

235.8

iqr

150.2

skew

7.511

kurtosis

77.6

n_outliers

159

outlier_rate

0.04915

zero_rate

0

This column names the nearest mineral deposit for each record, with 97 distinct sites across 3,235 rows and no nulls. Distribution is moderately concentrated: "Hatchet Creek Copper" alone accounts for 13.4% (434 rows), and the top three deposits cover roughly 30% of the data, yet entropy ratio of 0.76 indicates the long tail still carries meaningful spread. Names mix mine types (copper, clay, sulfur), pits, banks, quads, and districts, suggesting heterogeneous source nomenclature rather than a clean controlled vocabulary. Treatment: Treat as a high-cardinality categorical: target- or frequency-encode, and consider grouping rare deposits into an 'other' bucket. high · anthropic:claude-opus-4-7

n

3,235

nulls

0 (0.0%)

unique

97

top_value

Hatchet Creek Copper

top_rate

0.1342

cardinality

97

entropy

4.999

entropy_ratio

0.7574

Categorical label identifying the type of mineral or fuel deposit, with 10 distinct values across 3235 rows and no nulls. Coal dominates at 41.6% (1345 rows), followed by Copper, Iron, and Oil, while Zinc (23) and Silver (21) are rare. Entropy ratio of 0.76 indicates a moderately concentrated distribution skewed toward fossil/base resources rather than precious metals. Treatment: One-hot encode; consider grouping rare classes (Zinc, Silver) if used as a target. high · anthropic:claude-opus-4-7

n

3,235

nulls

0 (0.0%)

unique

10

top_value

Coal

top_rate

0.4158

cardinality

10

entropy

2.536

entropy_ratio

0.7633

Categorical geological era/period label for deposits, spanning 9 distinct values across 3235 complete rows. Distribution is unusually flat for a categorical (entropy_ratio 0.945) — Pennsylvanian leads at only 22.6% (732 rows) and even the smallest, Permian, holds 95 rows. Note the mixed granularity: broad eras (Paleozoic, Precambrian) sit alongside specific periods (Devonian, Miocene), so categories are not mutually exclusive in geological time. Treatment: One-hot encode, but consider reconciling overlapping era/period granularity before modelling. high · anthropic:claude-opus-4-7

n

3,235

nulls

0 (0.0%)

unique

9

top_value

Pennsylvanian

top_rate

0.2263

cardinality

9

entropy

2.997

entropy_ratio

0.9453

`deposit_state` is a categorical US-state field with 25 distinct values across 3,235 rows and no nulls. Distribution is fairly even (entropy ratio 0.83); the top state Missouri accounts for only 14.8%, followed closely by Ohio (448) and Alabama (434). Coverage is partial — only half the US states appear — so this is not a nationwide sample. Treatment: One-hot or target-encode for modelling; verify whether the 25-state coverage is intentional. high · anthropic:claude-opus-4-7

n

3,235

nulls

0 (0.0%)

unique

25

top_value

Missouri

top_rate

0.1478

cardinality

25

entropy

3.85

entropy_ratio

0.829