saturn·

← all readings

economic unemployment by county

source /home/coolhand/datasets/us-inequality-atlas/economic/unemployment_by_county.csv 3,222 rows 8 columns profiled 2026-05-01 raw JSON static .html .ipynb Report Notebook

Reading

dataset summary · high confidence anthropic:claude-opus-4-7

This dataset contains 3,222 US county-level labor market records with 8 columns covering county identifiers (FIPS, name, state) and workforce statistics (labor force, total 16+, unemployed, unemployment rate, participation rate). The unemployment rate averages 5.13% with a median of 4.69%, ranging up to 31.99%, so the right tail is worth inspecting for distressed counties. Population-based counts (labor_force, total_16_plus, unemployed) are extremely right-skewed (skew >13) with hundreds of outliers — expected when a few large metros sit alongside small rural counties, but it means you should likely log-transform before modeling. Texas (254), Georgia (159), and Virginia (133) contribute the most counties, reflecting state geography rather than any sampling bias. County names show a 39% duplicate rate driven by repeated names like Washington, Jefferson, and Franklin Counties across states — join on FIPS, not name.

citing: row_count · column_count · columns.unemployment_rate.stats · columns.labor_force.stats · columns.total_16_plus.stats · columns.unemployed.stats · columns.labor_force_participation_rate.stats · columns.state.top_values · columns.county_name.stats · columns.county_name.top_values

unemployment_rate · Check the right tail beyond ~10% to find counties with persistent labor distress.
Show data table
bincount
0 – 0.799760
0.7997 – 1.59990
1.599 – 2.399197
2.399 – 3.199333
3.199 – 3.999464
3.999 – 4.798539
4.798 – 5.598492
5.598 – 6.398361
6.398 – 7.198217
7.198 – 7.997142
7.997 – 8.79785
8.797 – 9.59761
9.597 – 10.436
10.4 – 11.236
11.2 – 1213
12 – 12.818
12.8 – 13.613
13.6 – 14.412
14.4 – 15.211
15.2 – 15.998
15.99 – 16.795
16.79 – 17.594
17.59 – 18.392
18.39 – 19.191
19.19 – 19.993
19.99 – 20.793
20.79 – 21.594
21.59 – 22.393
22.39 – 23.192
23.19 – 23.993
23.99 – 24.791
24.79 – 25.590
25.59 – 26.390
26.39 – 27.190
27.19 – 27.990
27.99 – 28.790
28.79 – 29.590
29.59 – 30.390
30.39 – 31.191
31.19 – 31.992
labor_force_participation_rate · Roughly symmetric around 58% — outliers below 30% flag counties with weak workforce engagement.
Show data table
bincount
18.63 – 20.271
20.27 – 21.90
21.9 – 23.540
23.54 – 25.171
25.17 – 26.811
26.81 – 28.441
28.44 – 30.080
30.08 – 31.714
31.71 – 33.356
33.35 – 34.988
34.98 – 36.6210
36.62 – 38.2524
38.25 – 39.8930
39.89 – 41.5237
41.52 – 43.1651
43.16 – 44.7961
44.79 – 46.4360
46.43 – 48.0676
48.06 – 49.7109
49.7 – 51.34141
51.34 – 52.97186
52.97 – 54.61174
54.61 – 56.24235
56.24 – 57.88245
57.88 – 59.51272
59.51 – 61.15270
61.15 – 62.78277
62.78 – 64.42252
64.42 – 66.05221
66.05 – 67.69187
67.69 – 69.32118
69.32 – 70.9682
70.96 – 72.5941
72.59 – 74.2319
74.23 – 75.8610
75.86 – 77.54
77.5 – 79.136
79.13 – 80.771
80.77 – 82.40
82.4 – 84.041
state · Counts per state show TX, GA, and VA dominate the row count simply due to having more counties.
Show data table
valuecountshare
TX2547.9%
GA1594.9%
VA1334.1%
KY1203.7%
MO1153.6%
KS1053.3%
IL1023.2%
NC1003.1%
IA993.1%
TN952.9%
NE932.9%
IN922.9%
OH882.7%
MN872.7%
MI832.6%
MS822.5%
PR782.4%
OK772.4%
AR752.3%
WI722.2%
labor_force · Severe right skew (max 5.2M vs median 11.6K) — consider a log scale before any modeling.
Show data table
bincount
36 – 1.311e+052944
1.311e+05 – 2.621e+05136
2.621e+05 – 3.931e+0552
3.931e+05 – 5.241e+0541
5.241e+05 – 6.551e+0514
6.551e+05 – 7.862e+0510
7.862e+05 – 9.172e+055
9.172e+05 – 1.048e+065
1.048e+06 – 1.179e+064
1.179e+06 – 1.31e+062
1.31e+06 – 1.441e+063
1.441e+06 – 1.572e+060
1.572e+06 – 1.703e+061
1.703e+06 – 1.834e+061
1.834e+06 – 1.965e+060
1.965e+06 – 2.096e+060
2.096e+06 – 2.227e+060
2.227e+06 – 2.358e+061
2.358e+06 – 2.489e+061
2.489e+06 – 2.62e+060
2.62e+06 – 2.751e+060
2.751e+06 – 2.882e+061
2.882e+06 – 3.013e+060
3.013e+06 – 3.145e+060
3.145e+06 – 3.276e+060
3.276e+06 – 3.407e+060
3.407e+06 – 3.538e+060
3.538e+06 – 3.669e+060
3.669e+06 – 3.8e+060
3.8e+06 – 3.931e+060
3.931e+06 – 4.062e+060
4.062e+06 – 4.193e+060
4.193e+06 – 4.324e+060
4.324e+06 – 4.455e+060
4.455e+06 – 4.586e+060
4.586e+06 – 4.717e+060
4.717e+06 – 4.848e+060
4.848e+06 – 4.979e+060
4.979e+06 – 5.11e+060
5.11e+06 – 5.241e+061
unemployed · Heavy-tailed with 417 outliers; a handful of large counties account for most absolute unemployment.
Show data table
bincount
0 – 91393018
9139 – 1.828e+0496
1.828e+04 – 2.742e+0453
2.742e+04 – 3.655e+0423
3.655e+04 – 4.569e+048
4.569e+04 – 5.483e+044
5.483e+04 – 6.397e+043
6.397e+04 – 7.311e+044
7.311e+04 – 8.225e+044
8.225e+04 – 9.139e+043
9.139e+04 – 1.005e+051
1.005e+05 – 1.097e+052
1.097e+05 – 1.188e+050
1.188e+05 – 1.279e+050
1.279e+05 – 1.371e+050
1.371e+05 – 1.462e+050
1.462e+05 – 1.554e+050
1.554e+05 – 1.645e+051
1.645e+05 – 1.736e+050
1.736e+05 – 1.828e+050
1.828e+05 – 1.919e+050
1.919e+05 – 2.01e+051
2.01e+05 – 2.102e+050
2.102e+05 – 2.193e+050
2.193e+05 – 2.285e+050
2.285e+05 – 2.376e+050
2.376e+05 – 2.467e+050
2.467e+05 – 2.559e+050
2.559e+05 – 2.65e+050
2.65e+05 – 2.742e+050
2.742e+05 – 2.833e+050
2.833e+05 – 2.924e+050
2.924e+05 – 3.016e+050
3.016e+05 – 3.107e+050
3.107e+05 – 3.199e+050
3.199e+05 – 3.29e+050
3.29e+05 – 3.381e+050
3.381e+05 – 3.473e+050
3.473e+05 – 3.564e+050
3.564e+05 – 3.655e+051

Schema

8 columns
Alerts
fips numeric 0.0% 3,222
county_name text 0.0% 1,960
short_text duplicates
state categorical 0.0% 52
total_16_plus numeric 0.0% 3,148
high_skew outliers
labor_force numeric 0.0% 3,099
high_skew outliers
unemployed numeric 0.0% 1,859
high_skew outliers
labor_force_participation_rate numeric 0.0% 1,944
unemployment_rate numeric 0.0% 950
high_skew

fips

numeric identifier
This is the U.S. county FIPS code, evidenced by every one of the 3222 rows being unique with no nulls and values spanning 1001 to 72153 — the standard 5-digit state+county encoding. The distribution is essentially uniform across the FIPS range (skew 0.16, kurtosis -0.63, no outliers), which is expected for an identifier rather than a measured quantity. Treatment: Treat as a categorical county key; left-join on this id rather than feeding it as a numeric feature. high · anthropic:claude-opus-4-7
n
3,222
nulls
0 (0.0%)
unique
3,222
min
1,001
max
72,153
mean
3.138e+04
median
30,022
std
1.63e+04
q1
1.903e+04
q3
4.61e+04
iqr
27,075
skew
0.1574
kurtosis
-0.6314
n_outliers
0
outlier_rate
0
zero_rate
0

county_name

text metadata short_text duplicates
This column holds US county-level place names: 3,222 rows with 1,960 unique values, all between 10 and 46 characters and averaging ~2 words. The vocabulary is dominated by 'county' (2,999 occurrences) but also includes 'municipio' (78, Puerto Rico), 'parish' (64, Louisiana), and 'city' (47), so the field mixes several jurisdiction types. Note the 39.2% duplicate rate — recurring names like Washington County (30), Jefferson County (25), and Franklin County (24) appear across many states, so this name alone does not uniquely identify a county. Treatment: Pair with a state code to form a unique key before joining or aggregating. high · anthropic:claude-opus-4-7
n
3,222
nulls
0 (0.0%)
unique
1,960
len_min
10
len_max
46
len_mean
14.17
len_median
14
len_p95
18
word_mean
2.083
word_median
2
n_empty
0
n_duplicates
1,262
duplicate_rate
0.3917
vocab_size
1,963
readability_flesch_mean
33.36
emoji_rate
0
url_rate
0
one_word_rate
0
allcaps_rate
0
boilerplate_rate
0

state

categorical feature
Two-letter US state/territory codes across 3,222 rows with 52 distinct values and no nulls — consistent with the 50 states plus DC and likely Puerto Rico. Distribution tracks county counts rather than population: TX leads at 254 (7.88%), followed by GA (159), VA (133) and KY (120), suggesting one row per county/equivalent. High entropy ratio (0.93) confirms a fairly even spread across states. Treatment: one-hot or target-encode for modelling; useful as a join key to state-level reference tables. high · anthropic:claude-opus-4-7
n
3,222
nulls
0 (0.0%)
unique
52
top_value
TX
top_rate
0.07883
cardinality
52
entropy
5.314
entropy_ratio
0.9322

total_16_plus

numeric feature high_skew outliers
This is a numeric population-style count of people aged 16+, with 3222 non-null rows and 3148 unique values spanning 50 to 8,086,852. The distribution is severely right-skewed (skew 13.49, kurtosis 305.88): the median is 21,167.5 but the mean is 83,549.93 and 13.7% of rows (443) flag as outliers. The std of 265,514 dwarfs the IQR of 45,507.75, consistent with a long upper tail typical of geographic aggregates. Treatment: log-transform before regression to tame the heavy right tail. high · anthropic:claude-opus-4-7
n
3,222
nulls
0 (0.0%)
unique
3,148
min
50
max
8.087e+06
mean
8.355e+04
median
2.117e+04
std
2.655e+05
q1
8986
q3
5.449e+04
iqr
4.551e+04
skew
13.49
kurtosis
305.9
n_outliers
443
outlier_rate
0.1375
zero_rate
0

labor_force

numeric feature high_skew outliers
This column appears to be the size of the labor force per record, likely at a US county or similar geographic unit given the 3,222 rows and 3,099 unique values. The distribution is severely right-skewed (skew 13.29, kurtosis 295.22) with a median of 11,608.5 but a max of 5,240,842, and 14.2% of values flagged as outliers. No nulls or zeros, but the gap between Q3 (31,930.5) and the maximum signals a long tail of very large jurisdictions. Treatment: log-transform before modelling to tame the heavy right tail. high · anthropic:claude-opus-4-7
n
3,222
nulls
0 (0.0%)
unique
3,099
min
36
max
5.241e+06
mean
5.287e+04
median
1.161e+04
std
1.742e+05
q1
4777
q3
3.193e+04
iqr
2.715e+04
skew
13.29
kurtosis
295.2
n_outliers
459
outlier_rate
0.1425
zero_rate
0

unemployed

numeric feature high_skew outliers
This is a numeric count of unemployed persons per record, with 3222 rows, no nulls, and 1859 unique values. The distribution is severely right-skewed (skew 16.82, kurtosis 450.4): the median is 589 while the mean is 2827 and the max reaches 365544, and 417 rows (12.9%) flag as outliers. Only 0.56% of records are zero, so sparsity is not the issue—a few extreme values are. Treatment: Log-transform (or winsorize) before any modelling to tame the heavy right tail. high · anthropic:claude-opus-4-7
n
3,222
nulls
0 (0.0%)
unique
1,859
min
0
max
365,544
mean
2827
median
589
std
1.083e+04
q1
223
q3
1706
iqr
1482
skew
16.82
kurtosis
450.4
n_outliers
417
outlier_rate
0.1294
zero_rate
0.005587

labor_force_participation_rate

numeric feature
Numeric column capturing labor force participation rate, almost certainly expressed as a percentage given the 18.63 to 84.04 range and mean of 57.89. The distribution is moderately left-skewed (-0.58) with a tight IQR of 10.695 around a median of 58.72, and only 1.18% of values flagged as outliers. No nulls or zeros across 3,222 rows, and 1,944 unique values suggest a continuous measurement rather than a coded category. Treatment: Use as-is for modelling; optionally standardize since values are bounded percentages. high · anthropic:claude-opus-4-7
n
3,222
nulls
0 (0.0%)
unique
1,944
min
18.63
max
84.04
mean
57.89
median
58.72
std
8.041
q1
52.97
q3
63.66
iqr
10.7
skew
-0.5766
kurtosis
0.4502
n_outliers
38
outlier_rate
0.01179
zero_rate
0

unemployment_rate

numeric feature high_skew
This is a county- or region-level unemployment rate expressed as a percentage, with values from 0.0 to 31.99 and a median of 4.69. The distribution is heavily right-skewed (skew 2.55, kurtosis 12.81) with 154 outliers (4.78%) pulling the mean above the median, and a small share of zero readings (0.56%). Treatment: Apply a log or winsorising transform before regression to tame the right tail. high · anthropic:claude-opus-4-7
n
3,222
nulls
0 (0.0%)
unique
950
min
0
max
31.99
mean
5.127
median
4.69
std
2.926
q1
3.42
q3
6.08
iqr
2.66
skew
2.545
kurtosis
12.81
n_outliers
154
outlier_rate
0.0478
zero_rate
0.005587