language data glottolog languoid

Fixed 8-character single-token codes (e.g., 'melk1240', 'yang1299'), unique across all 23,740 rows with no nulls or duplicates. The pattern of four letters followed by four digits is consistent with Glottolog-style language identifiers, making this a primary key rather than analyzable text. Treatment: Use as the row key for joins; exclude from modelling features. high · anthropic:claude-opus-4-7

n

23,740

nulls

0 (0.0%)

unique

23,740

len_min

8

len_max

8

len_mean

8

len_median

8

len_p95

8

word_mean

1

word_median

1

n_empty

0

n_duplicates

0

duplicate_rate

0

vocab_size

20,000

readability_flesch_mean

86.11

emoji_rate

0

url_rate

0

one_word_rate

1

allcaps_rate

0

boilerplate_rate

0

Looks like a Glottolog-style language family identifier (e.g. 'atla1278' for Atlantic-Congo, 'aust1307' for Austronesian) tagging each of the 23,740 rows. The distribution is heavily skewed: the top family alone covers 20.0% of rows and the top 10 of 287 families dominate, yielding an entropy ratio of 0.598. Null rate is low at 1.81%. Treatment: left-join on this id to a language-family reference table; consider grouping the long tail before modelling. high · anthropic:claude-opus-4-7

n

23,740

nulls

429 (1.8%)

unique

287

top_value

atla1278

top_rate

0.2

cardinality

287

entropy

4.886

entropy_ratio

0.5984

Fixed-width 8-character single-token codes (e.g. 'book1242', 'uncl1493') with len_min=len_max=8 and one_word_rate=1.0 — these look like Glottolog-style language/family identifiers used as a parent reference. With 7338 unique values across 23740 rows and a 68.5% duplicate_rate, many children share parents; 'book1242' alone accounts for 399 rows. Null rate is low at 1.81%. Treatment: left-join on this id to a parent/language lookup table. high · anthropic:claude-opus-4-7

n

23,740

nulls

429 (1.8%)

unique

7,338

len_min

8

len_max

8

len_mean

8

len_median

8

len_p95

8

word_mean

1

word_median

1

n_empty

0

n_duplicates

15,973

duplicate_rate

0.6852

vocab_size

7,189

readability_flesch_mean

91.19

emoji_rate

0

url_rate

0

one_word_rate

1

allcaps_rate

0

boilerplate_rate

0

The `name` column holds 23,740 fully unique short labels (null_rate 0.0, n_unique equals n), with a mean length of 9.95 characters and 69.5% being a single word. Top tokens like `nuclear`, `central`, `western`, `eastern`, `northern`, `southern`, and `language` suggest these are entity/topic names rather than person names. Vocabulary is broad (17,915 distinct words across only ~1.4 words per row), and there are no duplicates, URLs, or emoji. Treatment: Treat as a unique label key; drop from modelling features or hash/embed if semantic content is needed. high · anthropic:claude-opus-4-7

n

23,740

nulls

0 (0.0%)

unique

23,740

len_min

1

len_max

58

len_mean

9.95

len_median

8

len_p95

22

word_mean

1.398

word_median

1

n_empty

0

n_duplicates

0

duplicate_rate

0

vocab_size

17,915

readability_flesch_mean

42.62

emoji_rate

0

url_rate

0

one_word_rate

0.6953

allcaps_rate

0.0001685

boilerplate_rate

0

Binary boolean flag indicating whether a record involves bookkeeping, with only two values across 23740 rows and no nulls. The distribution is severely imbalanced: 'False' covers 98.3% (23341 rows) versus only 399 'True' cases, yielding an entropy ratio of just 0.12. Treatment: Encode as boolean and apply class-imbalance handling (e.g., stratification or reweighting) if used as a target. high · anthropic:claude-opus-4-7

n

23,740

nulls

0 (0.0%)

unique

2

top_value

False

top_rate

0.9832

cardinality

2

entropy

0.1231

entropy_ratio

0.1231

This is a categorical taxonomy tag with exactly 3 levels (dialect, language, family) and no nulls across 23,740 rows. The distribution is well-spread (entropy_ratio 0.94), with 'dialect' leading at 46.0%, followed by 'language' (8,481) and 'family' (4,339). Looks like a linguistic classification level rather than a free-form attribute. Treatment: one-hot encode for modelling or use directly as a stratification key. high · anthropic:claude-opus-4-7

n

23,740

nulls

0 (0.0%)

unique

3

top_value

dialect

top_rate

0.46

cardinality

3

entropy

1.494

entropy_ratio

0.9426

This is a categorical status column with 6 levels matching UNESCO-style language endangerment categories (safe, definitely endangered, vulnerable, extinct, critically endangered, severely endangered). The distribution is heavily imbalanced: 'safe' accounts for 79.9% of 23,740 rows, while the rarest level 'severely endangered' has only 413 records. Entropy ratio is 0.44, confirming low diversity despite 6 classes. Treatment: Treat as ordinal target; stratify or rebalance before classification given the 80/20 dominance of 'safe'. high · anthropic:claude-opus-4-7

n

23,740

nulls

0 (0.0%)

unique

6

top_value

safe

top_rate

0.7989

cardinality

6

entropy

1.15

entropy_ratio

0.4447

Geographic latitude coordinate spanning -55.27 to 73.14, consistent with valid Earth latitudes. Two-thirds of rows are null (null_rate 0.6654), which severely limits coverage. The distribution is mildly right-skewed (0.54) with median 6.31 and IQR ~24.5, suggesting a bias toward northern-hemisphere but tropical-leaning observations. Treatment: Impute or filter the 66% nulls before any spatial modelling; pair with longitude for geo-features. high · anthropic:claude-opus-4-7

n

23,740

nulls

15,797 (66.5%)

unique

7,798

min

-55.27

max

73.14

mean

8.17

median

6.306

std

18.96

q1

-5.137

q3

19.34

iqr

24.47

skew

0.5403

kurtosis

0.3006

n_outliers

129

outlier_rate

0.01624

zero_rate

0

Geographic longitude in decimal degrees, with values spanning -178.785 to 179.306, consistent with the global WGS84 range. Two-thirds of rows are null (null_rate 0.6654), so coverage is the dominant concern; among populated rows the distribution is mildly left-skewed (-0.48) with a median of 47.72 suggesting an Eastern-Hemisphere bias. Only 13 outliers (0.16%) and no zeros, so the populated values themselves look clean. Treatment: Pair with latitude for geospatial features; impute or filter the 66.5% missing before modelling. high · anthropic:claude-opus-4-7

n

23,740

nulls

15,797 (66.5%)

unique

7,757

min

-178.8

max

179.3

mean

51.27

median

47.72

std

81.14

q1

7.235

q3

124.1

iqr

116.9

skew

-0.4832

kurtosis

-0.7745

n_outliers

13

outlier_rate

0.001637

zero_rate

0

This column holds ISO 639-3 language codes: every non-null value is exactly 3 characters and one word (len_min=len_max=3, one_word_rate=1.0), with 7968 distinct codes across 23740 rows. Two-thirds are missing (null_rate=0.6644), and the cardinality is near-unique among populated rows, suggesting one code per language entry rather than a repeated categorical. Treatment: Treat as a language-code key; left-join to an ISO 639-3 reference table and handle the 66% nulls explicitly. high · anthropic:claude-opus-4-7

n

23,740

nulls

15,772 (66.4%)

unique

7,968

len_min

3

len_max

3

len_mean

3

len_median

3

len_p95

3

word_mean

1

word_median

1

n_empty

0

n_duplicates

0

duplicate_rate

0

vocab_size

7,968

readability_flesch_mean

119.5

emoji_rate

0

url_rate

0

one_word_rate

1

allcaps_rate

0

boilerplate_rate

0

This column is named 'description' but saturn skipped profiling, so no kind, uniqueness, or value statistics are available. Only the row count (23740) and a null rate of 0.0 are reported. Without further stats, the content and structure cannot be characterized. Treatment: Re-profile or sample manually before deciding; if textual, tokenize and embed before modelling. low · anthropic:claude-opus-4-7

n

23,740

nulls

0 (0.0%)

unique

—

This column was skipped by the profiler, so no statistics, uniqueness count, or value samples are available beyond a row count of 23,740 and a null rate of 0.0. The name suggests it holds markup or descriptive text (likely HTML or formatted product/item descriptions), but that is inferred from the label alone, not from evidence. No distributional signal can be assessed here. Treatment: Re-run the profiler with text handling enabled, then tokenize and embed before modelling. low · anthropic:claude-opus-4-7

n

23,740

nulls

0 (0.0%)

unique

—

A numeric count of child families per record, with 23740 rows and only 88 distinct values. It is overwhelmingly zero (zero_rate 0.9082) so q1, median, and q3 are all 0 and IQR is 0, yet a long tail pushes max to 859 with mean 0.879 and std 13.20. Skew of 44.40 and kurtosis of 2352.94 confirm an extreme heavy-tailed distribution, and 2179 rows (9.18%) flag as outliers. Treatment: Binarize (zero vs non-zero) or apply log1p before modelling to tame the extreme skew. high · anthropic:claude-opus-4-7

n

23,740

nulls

0 (0.0%)

unique

88

min

0

max

859

mean

0.8792

median

0

std

13.2

q1

0

q3

0

iqr

0

skew

44.4

kurtosis

2353

n_outliers

2,179

outlier_rate

0.09179

zero_rate

0.9082

A numeric count of child languages per record, where 81.7% of rows are zero and Q1=median=Q3=0, so the typical entity has none. The distribution is extremely long-tailed (skew 41.86, kurtosis 2115) with a max of 1435 and 4339 outliers (18.3% outlier rate), suggesting a small set of hub-like records dominate. Treatment: Binarize (has_children vs none) or log1p-transform before modelling given the heavy zero-inflation and skew. high · anthropic:claude-opus-4-7

n

23,740

nulls

0 (0.0%)

unique

126

min

0

max

1,435

mean

1.996

median

0

std

23.41

q1

0

q3

0

iqr

0

skew

41.86

kurtosis

2115

n_outliers

4,339

outlier_rate

0.1828

zero_rate

0.8172

This is a count of child dialects per record, dominated by zeros (zero_rate 0.7442) with a median and Q3 of 0/1 yet a max of 2369. Skew of 42.22 and kurtosis of 2159 confirm an extreme long tail, and 17.99% of rows flag as outliers. The mean of 3.39 is pulled far above the median, so any aggregate using it will mislead. Treatment: Log1p-transform or bin (zero / one / many) before modelling. high · anthropic:claude-opus-4-7

n

23,740

nulls

0 (0.0%)

unique

164

min

0

max

2,369

mean

3.389

median

0

std

36.8

q1

0

q3

1

iqr

1

skew

42.22

kurtosis

2159

n_outliers

4,272

outlier_rate

0.1799

zero_rate

0.7442

This column holds ISO-style country codes (PG, ID, NG, AU, IN…) as a categorical feature with 680 distinct values across 23,740 rows. Coverage is poor — 64.24% of rows are null — and the non-null distribution is broad (entropy 6.49, ratio 0.69) with Papua New Guinea leading at only 10.29%. The 680 distinct codes far exceed the ~250 real countries, suggesting multi-value concatenations or non-standard tokens behind the 'country_ids' plural. Treatment: Split multi-code entries, normalise to ISO-3166, and impute or flag the 64% missing before one-hot or target encoding. high · anthropic:claude-opus-4-7

n

23,740

nulls

15,250 (64.2%)

unique

680

top_value

PG

top_rate

0.1029

cardinality

680

entropy

6.493

entropy_ratio

0.6901