ml 32m tags

This is almost certainly a user identifier stored as an integer, with 15,848 unique values spread across roughly 2M rows (so ~126 rows per user on average). Values range from 22 to 162,279 with no nulls or zeros, and the distribution is near-symmetric (skew 0.12, kurtosis -0.26), consistent with sparsely allocated account IDs rather than a meaningful numeric quantity. The 157,830 flagged outliers (7.9%) are an artefact of treating IDs as numeric — they are not statistical anomalies in any business sense. Treatment: Treat as a categorical/foreign key for joins and grouping; do not use as a numeric feature. high · anthropic:claude-opus-4-7

n

2,000,072

nulls

0 (0.0%)

unique

15,848

min

22

max

162,279

mean

8.193e+04

median

78,213

std

3.811e+04

q1

68,413

q3

103,698

iqr

35,285

skew

0.1179

kurtosis

-0.2601

n_outliers

157,830

outlier_rate

0.07891

zero_rate

0

Integer movieId spanning 1 to 292629 across 2,000,072 rows with 51,323 unique values and no nulls. The cardinality and bounded integer range indicate a foreign key into a movies table rather than a measured quantity; summary statistics like mean (71893.26) and skew (0.82) reflect ID allocation, not a meaningful distribution. Treatment: left-join on this id to a movies dimension; do not treat as numeric feature. high · anthropic:claude-opus-4-7

n

2,000,072

nulls

0 (0.0%)

unique

51,323

min

1

max

292,629

mean

7.189e+04

median

52,328

std

7.48e+04

q1

4,011

q3

122,294

iqr

118,283

skew

0.821

kurtosis

-0.4024

n_outliers

0

outlier_rate

0

zero_rate

0

This column holds short descriptive tags applied to items (likely films), with frequent values like 'sci-fi', 'atmospheric', 'action', and 'comedy'. Entries are tiny — mean length 11.1 characters and word_mean 1.67 — and 52.5% are a single word. Despite 2,000,072 rows there are only 140,981 distinct tags and a 92.95% duplicate rate (1,859,091 duplicates), so the same labels recur heavily across items. Treatment: Treat as a categorical/multi-label tag field; one-hot or embed top tags and group the long tail. high · anthropic:claude-opus-4-7

n

2,000,072

nulls

0 (0.0%)

unique

140,981

len_min

1

len_max

241

len_mean

11.11

len_median

10

len_p95

21

word_mean

1.674

word_median

1

n_empty

0

n_duplicates

1.859e+06

duplicate_rate

0.9295

vocab_size

7,273

readability_flesch_mean

17.7

emoji_rate

0

url_rate

1.3e-05

one_word_rate

0.5247

allcaps_rate

0.01055

boilerplate_rate

1e-05

Unix epoch timestamps spanning roughly 1135429210 (late 2005) to 1697154983 (late 2023), with a median of 1574071062 (late 2019). The distribution is left-skewed (skew -1.22) and 6.39% of rows fall outside the IQR fence, indicating a long tail of older records before activity concentrated post-2018. With 1291250 unique values across 2000072 rows there is meaningful repetition but no nulls or zeros. Treatment: Convert from epoch seconds to datetime and derive features (year, month, recency); consider filtering or bucketing the pre-2018 long tail. high · anthropic:claude-opus-4-7

n

2,000,072

nulls

0 (0.0%)

unique

1,291,250

min

1.135e+09

max

1.697e+09

mean

1.529e+09

median

1.574e+09

std

1.291e+08

q1

1.474e+09

q3

1.615e+09

iqr

1.411e+08

skew

-1.219

kurtosis

0.6854

n_outliers

127,898

outlier_rate

0.06395

zero_rate

0