Clean label
Toyota CAMRY
40,832
hidden by splits
137,828
cleaned records
7
raw variants
29.6%
fragmented
Raw source labels
TOYT CAMRY
96,996
TOYOTA CAMRY
40,711
TOYO CAMRY
54
TOYTA CAMRY
24
TOYT - TOYOTA CAMRY
22
Data cleaning impactPublic CCRS analysis
The public record is only as useful as the labels behind it.
Crash records guide city meetings, newsroom charts, grant narratives, enforcement reviews, legal marketing, and daily route choices. When one vehicle or crash cause is split across typos, abbreviations, and swapped fields, the public sees a weaker signal than the record actually contains.
This page shows how much interpretation changes after standardization, while keeping the original CCRS source values available for audit.
Vehicle records reviewed
5,310,041
2,539,076 crash records represented
Make and model changed
2,172,968
40.9% of reviewed vehicle party records
Raw labels collapsed
10,133
90,583 raw labels to 80,450 cleaned labels
Active cleaning rules
329
Rules feeding the warehouse before public vehicle analysis
For residents
A safer-street conversation should not depend on whether one model was typed three different ways.
For city staff
Bad labels can move a problem down the list before it ever reaches a grant memo, council packet, or safety plan.
For newsrooms
A ranking built from raw labels can turn a typo pattern into a headline. Cleaning is part of responsible reporting.
For police leaders
Fragmented records make it harder to see where training, forms, or review workflows are creating public confusion.
Why this matters
A real pattern can disappear into spelling noise.
Raw labels are not neutral. If a model is recorded several ways, a basic "top vehicles" chart undercounts it. The corrected view groups those labels into one analytical name, then keeps the raw variants visible.
Current example: Toyota CAMRY recovers 40,832 records that would otherwise sit under smaller aliases.
Clean label
Honda ACCORD
39,094
hidden by splits
132,118
cleaned records
4
raw variants
29.6%
fragmented
Raw source labels
HOND ACCORD
93,024
HONDA ACCORD
39,067
HOND - HONDA ACCORD
25
HONDA` ACCORD
2
Clean label
Honda CIVIC
38,016
hidden by splits
138,222
cleaned records
3
raw variants
27.5%
fragmented
Raw source labels
HOND CIVIC
100,206
HONDA CIVIC
37,997
HOND - HONDA CIVIC
19
Clean label
Toyota COROLLA
30,000
hidden by splits
108,089
cleaned records
9
raw variants
27.8%
fragmented
Raw source labels
TOYT COROLLA
78,089
TOYOTA COROLLA
29,883
TOYO COROLLA
58
TOYTA COROLLA
22
TOYT - TOYOTA COROLLA
21
Before and after
The leaderboard can move after labels are grouped.
This is where data quality becomes a decision risk. A raw-label chart can rank a vehicle lower simply because its records were scattered across aliases.
| After cleaning | Best raw version | Recovered | Decision shift |
|---|---|---|---|
#2 Toyota CAMRY 159,424 cleaned records | TOYT CAMRY raw rank #3 | 53,145 | +1 places |
#3 Honda ACCORD 154,432 cleaned records | HOND ACCORD raw rank #2 | 47,870 | +-1 places |
#1 Honda CIVIC 161,687 cleaned records | HOND CIVIC raw rank #1 | 46,873 | same rank |
#4 Toyota COROLLA 125,263 cleaned records | TOYT COROLLA raw rank #4 | 39,812 | same rank |
#15 Tesla Model 3 33,642 cleaned records | TESL MODEL 3 raw rank #97 | 26,365 | +82 places |
#6 Toyota TACOMA 76,547 cleaned records | TOYT TACOMA raw rank #7 | 23,289 | +1 places |
#5 Chevrolet SILVERADO 86,193 cleaned records | CHEV SILVERADO raw rank #6 | 22,541 | +1 places |
#7 Toyota PRIUS 70,186 cleaned records | TOYT PRIUS raw rank #8 | 20,796 | +1 places |
#22 Tesla Model Y 27,818 cleaned records | TESL Y raw rank #96 | 20,512 | +74 places |
#9 Nissan ALTIMA 66,746 cleaned records | NISS ALTIMA raw rank #11 | 20,421 | +2 places |
Where the answer is fragile
ZIPs most sensitive to raw labels.
These ZIP areas have the highest share of changed vehicle labels. They deserve extra caution before someone uses raw records to describe local crash patterns.
Unincorporated / 95387
848 of 1,081 vehicle records changed
78.4%
Freightliner CASCADIA
47
Honda CIVIC
35
Toyota CAMRY
34
Unincorporated / 94128
1,058 of 1,351 vehicle records changed
78.3%
Toyota COROLLA
56
Toyota PRIUS
52
Honda ACCORD
50
Unincorporated / 93653
125 of 160 vehicle records changed
78.1%
Chevrolet SILVERADO
10
Toyota COROLLA
8
Honda ACCORD
4
Unincorporated / 95527
102 of 131 vehicle records changed
77.9%
Toyota CAMRY
8
Chevrolet SILVERADO
5
Toyota HIGHLANDER
5
Hayward / 94587
3,216 of 4,149 vehicle records changed
77.5%
Toyota CAMRY
180
Honda ACCORD
169
Honda CIVIC
167
Current highest-impact ZIP: Unincorporated / 95387.
Reporting quality
Messy labels become public blind spots.
This is not a blame scoreboard. It is an audit trail for the places where public crash records need the most translation before anyone should compare them.
A fragmented record can make a vehicle trend, crash cause, or neighborhood risk look smaller than it is. That affects public meetings, newsroom charts, enforcement reviews, grant narratives, and the everyday driver trying to understand local danger.
Jurisdiction currently uses the CCRS city label as a public reporting proxy until a dedicated police-department dimension is exposed.
Places where vehicle labels changed most
Minimum 250 vehicle party records. Higher share means raw labels are more likely to distort local comparisons.
Seal Beach
3,689 changed vehicle labels, 882 raw values collapsed to 806
75.1%
vehicle labels changed
Lafayette
1,876 changed vehicle labels, 632 raw values collapsed to 570
71.5%
vehicle labels changed
Unincorporated
1,016,931 changed vehicle labels, 46,049 raw values collapsed to 41,802
71.0%
vehicle labels changed
Orinda
1,406 changed vehicle labels, 528 raw values collapsed to 475
70.6%
vehicle labels changed
South San Francisco
3,825 changed vehicle labels, 1,208 raw values collapsed to 1,091
69.7%
vehicle labels changed
Menlo Park
3,039 changed vehicle labels, 1,024 raw values collapsed to 952
69.3%
vehicle labels changed
Newport Beach
5,397 changed vehicle labels, 1,600 raw values collapsed to 1,541
68.8%
vehicle labels changed
Delano
1,904 changed vehicle labels, 571 raw values collapsed to 524
68.5%
vehicle labels changed
Year-by-year cleanup sensitivity
If this line stays high, the issue is not a one-year typo batch.
Lafayette
2020 to 2026
20
21
22
23
24
25
26
Orinda
2020 to 2026
20
21
22
23
24
25
26
Seal Beach
2020 to 2026
20
21
22
23
24
25
26
South San Francisco
2020 to 2026
20
21
22
23
24
25
26
Unincorporated
2020 to 2026
20
21
22
23
24
25
26
Crash-cause cleanup not shown yet
Primary Collision Factor means the reported crash cause or violation. Those mappings are still being reconciled, so this page avoids showing a misleading 0.0% score.
Cleaning progress
The current rulebook.
These mappings are applied before the public pages calculate vehicle rankings. Raw source values are preserved, so every cleaned label can still be audited.
Vehicle Make
77 clean labels
210
rules
Vehicle Make V2
38 clean labels
68
rules
Vehicle Model
9 clean labels
51
rules
Audit examples
Raw values mapped into clean labels.
Model 3
14 raw values
Vehicle Model
MAZD || MODEL 3TELSA || MODEL3TESL || MODEL 3TESLA MOTORS || MODEL 3TESLA || 3TESLA || Model3
Model Y
13 raw values
Vehicle Model
TELSA || MODEL YTESL || MODEL-YTESL || MODELY YTESLA / || MODEL Y /TESLA MOTORS || MODEL YTESLA || Y
Tesla
12 raw values
Vehicle Make
TELSATESLTESL - TESLATESLA /TESLA MOTORSTESLA-TSMR
Model S
9 raw values
Vehicle Model
TELSA || MODEL STESL || model sTESLA MOTORS || Model STESLA || MODel sTESLA || s modelTSMR || MODEL S
Model X
8 raw values
Vehicle Model
BMW || X-ModelsTELSA || MODEL XTESL || Model XTESLA MOTORS || MODEL XTESLA || model xTSMR || modelx
Blue Bird
7 raw values
Vehicle Make
BLBRDBLUBLUBBLUEBLUEBBLUEBIRD
comparison
Compares source vehicle labels to the current cleaned warehouse labels.
privacy
The public report uses aggregate counts only and does not expose individual crash records.
scope
Summary totals include all vehicle labels. Example rankings filter out Unknown model labels so the public story focuses on resolved make-model cleanup.
zip threshold
ZIP impact examples require at least 100 party vehicle records in the selected window.
jurisdiction
Reporting-jurisdiction examples use CCRS city labels as the public agency proxy because no separate police-department dimension is currently exposed in the mart.
Source: California Highway Patrol CCRS public records, processed independently in BigQuery. This page measures label standardization impact. It does not claim that one vehicle is more dangerous than another, and crash records are not adjusted for registrations, miles driven, or population exposure.