Spatial Visualization

2341181215756101120241416321998132117222537313427293244354748284326334639303641424538406467605468586152705962515363565765507155696649728293858477789688737581867495808397987987917692899411011110410610811511410711910599121109101103113122102112100116118120117129140146137144148160152161164168133131153145141134125136138132135155162171151123156143157170139142150169147126128154165158149124159167127163166176182173203184195213174198175178180183220194200206186202216221179172181190177188191201204209219189214215192217208212193196207211187185205218197199210235227222223229224239225232242244233230246228231245248236234238226241240243270271268251273259255266253249254256250258274267257264252272260261262263265278275285330299323293325328347350291294280287295320326335292300318319332340341334305297282276290296313333304331337346283286307309327344348349277311317324303339343302308310315321329284298279281312316288289306345301314322336351364367355356368370378391371393358363369400373376395361377354381353374380357385388383390387394362384359396352372375379382386389397398399401402360366392407404406408409405403410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511513514515516517518519520521522523524525526527528529530531532533534535536537538539540541543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645647648650651652653654655656659660662663664665668669671672673675676677678679680681682683684686687688689

The spatial view above shows each incident in the database as a plot point containing its incident ID number. Incidents are positioned so that those with similar report texts fall closer together. For example, incidents concerning autonomous vehicles form a tight cluster. We determine incident similarity using a natural language processing system, which you can read more about in our blog post on its rollout.