Data Report — Chronic Kidney Disease

SemMap JSON-LD: dataset.semmap.json · RDFa HTML

Overview

Metric	Value
Dataset	Chronic Kidney Disease
Source	UCI dataset 336
Rows	158
Columns	25
Discrete	13
Continuous	12
SemMap	SemMap JSON-LD SemMap HTML
Missingness	Not modeled

Variables and summary

variable	inferred	dist
age	continuous	49.5633 ± 15.5122 [6, 39.25, 50.5, 60, 83]
bp	continuous	74.0506 ± 11.1754 [50, 60, 80, 80, 110]
sg	continuous	1.0199 ± 0.0055 [1.005, 1.02, 1.02, 1.025, 1.025]
al	discrete	Albumin negative [0]: 116 (73.42%) Albumin 2+ [3]: 15 (9.49%) Albumin 3+ [4]: 15 (9.49%) Albumin 1+ [2]: 9 (5.70%) Trace albumin [1]: 3 (1.90%) Albumin 4+ [5]: 0 (0.00%)
su	discrete	Sugar negative [0]: 140 (88.61%) Trace sugar [1]: 6 (3.80%) Sugar 1+ [2]: 6 (3.80%) Sugar 2+ [3]: 3 (1.90%) Sugar 3+ [4]: 2 (1.27%) Sugar 4+ [5]: 1 (0.63%)
rbc	discrete	Normal [normal]: 140 (88.61%)
pc	discrete	Normal [normal]: 129 (81.65%)
pcc	discrete	Not present [notpresent]: 144 (91.14%)
ba	discrete	Not present [notpresent]: 146 (92.41%)
bgr	continuous	131.3418 ± 64.9398 [70, 97, 115.5, 131.75, 490]
bu	continuous	52.5759 ± 47.3954 [10, 26, 39.5, 49.75, 309]
sc	continuous	2.1886 ± 3.0776 [0.4, 0.7, 1.1, 1.6, 15.2]
sod	continuous	138.8481 ± 7.4894 [111, 135, 139, 144, 150]
pot	continuous	4.6367 ± 3.4764 [2.5, 3.7, 4.5, 4.9, 47]
hemo	continuous	13.6873 ± 2.8822 [3.1, 12.6, 14.25, 15.775, 17.8]
pcv	continuous	41.9177 ± 9.1052 [9, 37.5, 44, 48, 54]
wbcc	continuous	8475.9494 ± 3126.8802 [3800, 6525, 7800, 9775, 26400]
rbcc	continuous	4.8918 ± 1.0194 [2.1, 4.5, 4.95, 5.6, 8]
htn	discrete	Yes [yes]: 34 (21.52%)
dm	discrete	No [no]: 130 (82.28%) Yes [yes]: 28 (17.72%) No [no]: 0 (0.00%)
cad	discrete	Yes [yes]: 11 (6.96%)
appet	discrete	Good appetite [good]: 139 (87.97%)
pe	discrete	Yes [yes]: 20 (12.66%)
ane	discrete	Yes [yes]: 16 (10.13%)
class	discrete	Not chronic kidney disease [notckd]: 115 (72.78%) Chronic kidney disease [ckd]: 43 (27.22%) Chronic kidney disease [ckd]: 0 (0.00%)

Fidelity summary

model	backend	disc jsd mean	disc jsd median	cont ks mean	cont w1 mean	downstream sign match
metasyn	metasyn	0.0646	0.0514	0.216	56.8532	0.5263
clg_mi2	pybnesian	0.0744	0.0752	0.1996	62.539
semi_mi5	pybnesian	0.0744	0.0752	0.1996	62.539
ctgan_fast	synthcity	0.1733	0.1798	0.7529	1042.22
tvae_quick	synthcity	0.1744	0.193	0.2715	86.9495

Privacy summary

model	backend	n real	n synth	near duplicate rate eps	nn distance mean	k min	k pct lt5	k map	delta presence
metasyn	metasyn	158	400	0.9937	0.0165	1	1	1	2.5714
clg_mi2	pybnesian	158	400	0.9873	0.0271	1	1	1	3.5
semi_mi5	pybnesian	158	400	0.9873	0.0271	1	1	1	3.5
ctgan_fast	synthcity	158	256	0.3987	0.3484	1	1	1	28.6
tvae_quick	synthcity	158	256	0.9873	0.0241	1	1	1	11

Models

UMAP Details Structure

Real data

Model: metasyn (metasyn)

Per-variable fidelity

variable	type	KS	W1	JSD
age	continuous	0.1125	2.3701
bp	continuous	0.3	3.9054
sg	continuous	0.4062	0.0019
al	discrete			0.094
su	discrete			0.1747
rbc	discrete			0.0228
pc	discrete			0.0535
pcc	discrete			0.0514
ba	discrete			0.1342
bgr	continuous	0.2425	23.229

Downstream metrics

metric

value

sign_match_rate

0.5263

formula

col_class ~ Q('age') + Q('bp') + Q('sg') + Q('al') + Q('su') + C(Q('rbc'), levels=['normal', 'abnormal']) + C(Q('pc'), levels=['normal', 'abnormal']) + C(Q('pcc'), levels=['present', 'notpresent']) + C(Q('ba'), levels=['present', 'notpresent']) + Q('bgr') + Q('bu') + Q('sc') + Q('sod') + Q('pot') + Q('hemo') + Q('pcv') + Q('wbcc') + Q('rbcc') + C(Q('htn'), levels=['yes', 'no']) + C(Q('dm'), levels=['yes', 'no']) + C(Q('cad'), levels=['yes', 'no']) + C(Q('appet'), levels=['good', 'poor']) + C(Q('pe'), levels=['yes', 'no']) + C(Q('ane'), levels=['yes', 'no']) + Q('age'):Q('bp') + Q('bp'):Q('sg') + Q('sg'):Q('al') + Q('al'):Q('su') + Q('su'):C(Q('rbc'), levels=['normal', 'abnormal'])

Privacy metrics

metric	value
n_real	158
n_synth	400
exact_overlap_rate	0
near_duplicate_rate_eps	0.9937
nn_distance_mean	0.0165
k_min	1
k_pct_lt5	1
k_map	1
rare_qi_reproduction_rate	0
delta_presence	2.5714

variable	distribution
age	core.normal
bp	core.lognormal
sg	core.truncated_normal
al	core.multinoulli
su	core.multinoulli
rbc	core.multinoulli
pc	core.multinoulli
pcc	core.multinoulli
ba	core.multinoulli
bgr	core.truncated_normal
bu	core.lognormal
sc	core.lognormal
sod	core.truncated_normal
pot	core.normal
hemo	core.truncated_normal
pcv	core.truncated_normal
wbcc	core.lognormal
rbcc	core.normal
htn	core.multinoulli
dm	core.multinoulli
cad	core.multinoulli
appet	core.multinoulli
pe	core.multinoulli
ane	core.multinoulli
class	core.multinoulli

Model: clg_mi2 (pybnesian)

Seed: 42, rows: 400
Params: {"max_indegree": 2, "operators": ["arcs"], "score": "bic", "type": "clg"}
Synthetic CSV
Per-variable metrics
Metrics JSON
Privacy metrics

Per-variable fidelity

variable	type	KS	W1	JSD
age	continuous	0.1275	2.6951
bp	continuous	0.3025	3.9078
sg	continuous	0.3162	0.0019
al	discrete			0.0925
su	discrete			0.202
rbc	discrete			0.016
pc	discrete			0.095
pcc	discrete			0.0585
ba	discrete			0.0919
bgr	continuous	0.12	12.7667

Privacy metrics

metric	value
n_real	158
n_synth	400
exact_overlap_rate	0
near_duplicate_rate_eps	0.9873
nn_distance_mean	0.0271
k_min	1
k_pct_lt5	1
k_map	1
rare_qi_reproduction_rate	0
delta_presence	3.5

Model: semi_mi5 (pybnesian)

Seed: 42, rows: 400
Params: {"max_indegree": 5, "operators": ["arcs"], "score": "bic", "type": "semiparametric"}
Synthetic CSV
Per-variable metrics
Metrics JSON
Privacy metrics

Per-variable fidelity

variable	type	KS	W1	JSD
age	continuous	0.1275	2.6951
bp	continuous	0.3025	3.9078
sg	continuous	0.3162	0.0019
al	discrete			0.0925
su	discrete			0.202
rbc	discrete			0.016
pc	discrete			0.095
pcc	discrete			0.0585
ba	discrete			0.0919
bgr	continuous	0.12	12.7667

Privacy metrics

metric	value
n_real	158
n_synth	400
exact_overlap_rate	0
near_duplicate_rate_eps	0.9873
nn_distance_mean	0.0271
k_min	1
k_pct_lt5	1
k_map	1
rare_qi_reproduction_rate	0
delta_presence	3.5

Model: ctgan_fast (synthcity)

Seed: 42, rows: 256
Params: {"batch_size": 256, "n_iter": 5}
Synthetic CSV
Per-variable metrics
Metrics JSON
Privacy metrics

Per-variable fidelity

variable	type	KS	W1	JSD
age	continuous	1	34.0312
bp	continuous	0.1016	2.5
sg	continuous	0.2461	0.0028
al	discrete			0.3647
su	discrete			0.2342
rbc	discrete			0.1798
pc	discrete			0.2748
pcc	discrete			0.073
ba	discrete			0.1057
bgr	continuous	0.9062	178.988

Privacy metrics

metric	value
n_real	158
n_synth	256
exact_overlap_rate	0
near_duplicate_rate_eps	0.3987
nn_distance_mean	0.3484
k_min	1
k_pct_lt5	1
k_map	1
rare_qi_reproduction_rate	0
delta_presence	28.6

Model: tvae_quick (synthcity)

Seed: 42, rows: 256
Params: {"batch_size": 256}
Synthetic CSV
Per-variable metrics
Metrics JSON
Privacy metrics

Per-variable fidelity

variable	type	KS	W1	JSD
age	continuous	0.1562	4.4251
bp	continuous	0.1328	2.7344
sg	continuous	0.168	0.0018
al	discrete			0.2962
su	discrete			0.175
rbc	discrete			0.1031
pc	discrete			0.2671
pcc	discrete			0.073
ba	discrete			0.193
bgr	continuous	0.1992	13.2205

Privacy metrics

metric	value
n_real	158
n_synth	256
exact_overlap_rate	0
near_duplicate_rate_eps	0.9873
nn_distance_mean	0.0241
k_min	1
k_pct_lt5	1
k_map	1
rare_qi_reproduction_rate	0
delta_presence	11