Fast Company: With Mapbox Deal, IBM Watson Will Learn A Lot More About Where Things are Happening

By Sean Captain
August 10, 2016

With Map­box Deal, IBM Wat­son Will Learn A Lot More About Where Things Are Happening
The geo­da­ta provider will help IBM let non-techies track things like sales, real estate val­ues, and cit­i­zen com­plaints by address.

The catch­phrase “arti­fi­cial intel­li­gence” may be ubiq­ui­tous, but it’s hard to find peo­ple who actu­al­ly know how to use it. That dis­crep­an­cy rep­re­sents an oppor­tu­ni­ty for prod­ucts like Wat­son Ana­lyt­ics, IBM’s point-and-click AI tool that lets non-experts ana­lyze busi­ness data by ask­ing ques­tions in col­lo­qui­al speech. In recent years, IBM has been build­ing out the data sources that Wat­son Ana­lyt­ics can draw from—and is now adding gran­u­lar map­ping info through a deal with geo­da­ta provider Mapbox.

Wat­son Ana­lyt­ics is not the sex­i­est form of AI—pretty far from Ex Machi­na. But it’s attrac­tive to peo­ple in fields like mar­ket­ing, who can start with as lit­tle as a spread­sheet of sales fig­ures and get arti­fi­cial insights into how and where they might sell more prod­uct. Fast­Com­pa­ny(Sin­gle user accounts range from free to $80 per month.) Since announc­ing Wat­son Ana­lyt­ics in Sep­tem­ber 2014, IBM has been adding built-in data sources that cus­tomers can pair with their uploads. In March, 2015, it announced a deal with Twit­ter to let Wat­son Ana­lyt­ics users study online chat­ter. TFast­Com­pa­nyhat same month, IBM also made a deal for detailed mete­o­ro­log­i­cal data from The Weath­er Com­pa­ny (own­er of The Weath­er Chan­nel and Weath­er Underground)—including mov­ing its data to IBM’s cloud. A few months lat­er, IBM pur­chased The Weath­er Com­pa­ny, a deal that closed in Jan­u­ary 2016.
The deal adds more-detailed geo­da­ta to Wat­son Analytics
“For all these com­pa­nies, the sig­nal may not be in your own data,” says Marc Alt­shuller, IBM’s gen­er­al man­ag­er of busi­ness ana­lyt­ics. “Most like­ly, the sig­nal is in a com­bi­na­tion of mar­ket data like what Map­box has, plus your own inter­nal data.”

Cur­rent­ly, Wat­son Ana­lyt­ics’ geo­da­ta stops at the coun­ty lev­el; when Map­box is inte­grat­ed, it will go down to street address­es. (IBM has­n’t pro­vid­ed a time frame for the inte­gra­tion.) IBM expects this to open up oppor­tu­ni­ties for brick-and-mor­tar busi­ness­es to fig­ure out how sales vary by store loca­tion, for instance, or the best site for a new store based on fac­tors like traf­fic. With­out street-lev­el detail, “you may not see that it’s urban or rur­al that mat­ters, whether it’s in a shop­ping mall or out­side a shop­ping mall that mat­ters,” says Alt­shuller. “Maybe it’s some­thing relat­ed to traf­fic and time of that traffic.”
Oth­er poten­tial ways to use Map­box data: ana­lyz­ing nation­al sales by geo­graph­ic region, eval­u­at­ing the suc­cess of junk mail cam­paigns, or pro­ject­ing home val­ues based on fac­tors like crime and traf­fic. Maybe not excit­ing, but poten­tial­ly lucra­tive. None of these are new appli­ca­tions in the busi­ness world, but IBM is offer­ing the abil­i­ty for non-techies to do this kind of analy­sis for a few bucks a month—and with­out need­ing much tech­ni­cal know-how. Nat­ur­al lan­guage pro­cess­ing can return data on straight­for­ward typed ques­tions like: “What is the aver­age income on the Upper East Side of New York City?”

The deal will also inte­grate Map­box into anoth­er IBM prod­uct called Cog­nos Ana­lyt­ics, which is expect­ed to hap­pen by the end of the year. It’s a drag-and-drop inter­face to cre­ate reports or inter­ac­tive “dash­boards,” like a sin­gle screen that rep­re­sents data in bar graphs, line graphs, and tables. IBM has a lot of com­pe­ti­tion here, with rivals such as Microsoft Pow­er BI, Qlik, and Tableau.

IBM and Map­box cre­at­ed for us a few exam­ples of zoomable, inter­ac­tive maps for Cog­nos. One breaks down a week of 311 call com­plaints in San Fran­cis­co, with the abil­i­ty to drill down to each address. Anoth­er uses cen­sus data to explore how income and pop­u­la­tion vary by coun­ty around the U.S. While Wat­son and Cog­nos already had coun­ty-lev­el geo­graph­ic res­o­lu­tion, Map­box adds spiffi­er ways to visu­al­ize it, such as the zoom­ing capa­bil­i­ty, says IBM.
The deal is as much an enhance­ment of Wat­son and Cog­nos as an endorse­ment of the young map­ping com­pa­ny that’s tak­ing on behe­moths like Google and Apple. Found­ed in 2010, Map­box has raised about $62.5 mil­lion in the past three years. It pro­vides the map­ping com­po­nent for over 5,000 plat­forms, includ­ing, Etsy, Foursquare, Instacart, Lone­ly Plan­et, MapQuest, and Pin­ter­est. In June, the com­pa­ny launched a ser­vice called Map­box Dri­ve for pow­er­ing semi-autonomous vehicles.

CEO Eric Gun­der­sen has a back­ground in inter­na­tion­al devel­op­ment, and he was orig­i­nal­ly tar­get­ing clients like the UN and the World Bank through a con­sul­tan­cy com­pa­ny called Devel­op­ment Seed. “These are orga­ni­za­tions work­ing on the ground, fig­ur­ing out where to put clin­ics, map­ping malar­ia across Africa, [doing] elec­tion mon­i­tor­ing,” says Gundersen.

Bit by bit we had to build up not just the tools to visu­al­ize the data, but the under­ly­ing data set, the map of the world where all the cities are, the ter­rain and the satel­lite imagery,” he says. Map­box, which spun out from Devel­op­ment Seed, did this on the cheap, har­vest­ing data from Open­StreetMap, a non­prof­it, crowd­sourced map­ping project. Because Open­StreetMap is open-source, Map­box’s refine­ments are avail­able to every­one, and vice ver­sa. (The work was fund­ed by a $575,000 grant from the Knight Foun­da­tion.) It also pulls from oth­er pub­lic sources, such as USGS, Land­sat, Nat­ur­al Earth, and Ope­nAd­dress­es. In 2015, though, Map­box pur­chased the rights to 3 mil­lion square kilo­me­ters of imagery from Dig­i­tal­Globe, which cap­tures some of the high­est-res­o­lu­tion (non­mil­i­tary) satel­lite views available.
The com­pa­ny also makes Map­box Stu­dio, user-friend­ly map­mak­ing soft­ware, which got a thor­ough rework in 2015. It’s not the only such point-and-click tool. Car­to (for­mal­ly Car­toDB), for instance, pro­duces gor­geous geo­graph­ic visu­al­iza­tions of things like Twit­ter traf­fic dur­ing dif­fer­ent moments of the Acad­e­my Awards Car­to’s new geo­graph­ic analy­sis tool, called Builder, is a com­peti­tor to the new map­ping abil­i­ties that Cog­nos will get. (Map­box sup­plies data to Cog­nos com­peti­tors, as well, such as Tableau.)

Map­box was already work­ing out a deal to pro­vide data to The Weath­er Com­pa­ny, says Gun­der­sen, when IBM bought The Weath­er Com­pa­ny. A new update to the Weath­er Chan­nel iPhone app uses Map­box for cus­tomiz­able, zoomable time-lapse visu­als like show­ing how storms move across the Caribbean or how tem­per­a­tures fluc­tu­ate among city neigh­bor­hoods. The fea­ture is com­ing to the Android ver­sion of the Weath­er Chan­nel app, as well as to the Weath­er Under­ground app, says Gundersen.

IBM so val­ued weath­er data that it ulti­mate­ly bought the com­pa­ny that was pro­vid­ing it. Might it feel the same about The Weath­er Com­pa­ny’s erst­while busi­ness part­ner and snap up Map­box, as well? “We’re not for sale,” says Gun­der­sen. “It’s the ear­ly days of the entire space.”

Read more at : http://​www​.fast​com​pa​ny​.com/​3​0​6​2​6​3​5​/​w​i​t​h​-​m​a​p​b​o​x​-​d​e​a​l​-​i​b​m​-​w​a​t​s​o​n​-​w​i​l​l​-​k​n​o​w​-​w​h​e​r​e​-​t​h​i​n​g​s​-​a​r​e​-​h​a​p​p​e​n​ing