SpatioTemporal Analysis of Flight Delays Poster

Download SpatioTemporal Analysis of Flight Delays Poster

Post on 20-Feb-2017

72 views

Category:

Documents

1 download

Embed Size (px)

TRANSCRIPT

  • Air$Carrier$Delay$

    Extreme$Weather$Delay$

    Na2onal$Avia2on$

    System$Delay$

    Aircra9$Arriving$Late$

    Methods(Performed$on$airports$dataset$for$each$of$4$delay$type$aAributes$for$all$12$months:$ Global(Morans(I(test$for$spa2al$autocorrela2on$!

    o Inverse$distance$conceptualiza2on$of$spa2al$weights$$ Anselin(Locan(Morans(I(Clustering/Outlier$Analysis$(C/O)$ Ge5s6Ord(Gi*$Hot$Spot$Analysis$(HS)$

    Spa5al(and(Temporal(Pa>erns(of(Flight(Delay(Types(in(the(Con5guous(United(States(During(2012(

    Data(

    Sta5s5cally(significant(clustering(found(for:( Extreme(weather(delay$in$southern$Great$Plains$during$spring$and$summer$months$ NAS(delay(in$the$Rocky$Mountain$region$and$Minnesota/Michigan$area$during$winter$months$ NAS(delay(in$Florida$and$the$east$coast$during$summer$months$$ AircraP(arriving(late(delay(on$the$west$coast$of$the$United$States$during$all$months$

    Other(notable(trends:( NAS(delays$tend$to$be$more$common$(percentageUwise)$at$busier$airports$ Air(carrier(delays(showed$an$inverse$type$of$effect$in$rela2on$to$NAS(delays(in$terms$of$regions$that$experienced$clustering$$

    C/O$U$February$ C/O$U$August$

    C/O$U$December$

    Na2onal$Avia2on$System$Delay$

    Air$Carrier$Delay$

    HS$U$February$ HS$U$August$

    Objec5ve(To$understand$sta2s2cally$significant$temporal$and$spa2al$paAerns$of$four$

    reported$flight$delay$types$in$the$United$States$in$2012$

    $!$Do$certain$regions$

    exhibit$significant$clustering$of$higher$percentages$of$delay$types$during$certain$

    2mes$of$the$year?$

    Lauren$Anderson,$M.S.$Student$U$$University$of$Georgia,$Athens,$GA,$USA$$U$leanders@uga.edu$

    Introduc5on(Flight$ delays$ are$ a$ significant$ problem$ in$ the$United$ States,$with$ 20%$of$ all$ flights$ delayed$by$ 15$ minutes$ or$ more$ in$ 2012$ $ (22%$ in$2013)1,$cos2ng$billions$of$dollars$to$passengers$and$ airlines$ each$ year2.$ Understanding$ the$underlying$ paAerns$ in$ delay$ types$ related$ to$regional$ and$ seasonal$ weather$ condi2ons$throughout$the$country$is$an$important$step$in$gaining$ a$ complete$ understanding$ of$ the$problem$as$a$whole.$$Spa2al$sta2s2cs$allow$for$the$ detec2on$ of$ significant$ paAerns$ in$aAributes$ through$ space,$ and$ are$ u2lized$ in$this$ study$ in$ the$context$of$airports$and$ their$percentages$of$flight$delays$by$type.$$$

    !Are$there$significant!pa+erns?!

    !Where!are!the!significant!!!!!!!pa+erns?!

    Bureau$of$Transporta2on$Sta2s2cs$(BTS):$delay$data$o For$296$airports:$#$of$total$opera2ons$(listed$separately$for$each$air$

    carrier),$#$opera2ons$delayed,$#$opera2ons$delayed$by$type$$

    Openflights.org:$airports$shapefile$o 3UleAer$airport$code,$la2tude,$longitutde,$

    opera2onal$status$

    Note:$Figures$2U7$only$show$buffers$around$$airports$with$>$3500$departures$per$month.$$Buffer$size$=$(log([SUM_arr_fl])^3)*200.$The$buffers$are$displayed$to$help$visually$iden2fy$large$airports$(not$part$of$analysis).$$$

    ((((((Delay(Type(( %((((AircraP(Arriving(Late( 41.41($$$Air$Carrier$$ (54.45)$$$$Security$$ (0.22)$$$$Extreme$Weather$$ (6.86)$$$$NAS$ (38.47)$

    $$$Weather$ (70.03)$$$$Volume$ (19.13)$$$$Equipment$ (0.47)$$$$Closed$Runway$ (7.15)$$$$Other$ (3.22)$

    $$ $$(((Na5onal(Avia5on(((((((((System((NAS)( 22.54($$$$Weather$ (69.81)$$$$$Volume$ (19.23)$$$$$Equipment$ (0.45)$$$$$Closed$Runway$ (7.26)$$$$$Other$ (3.25)$

    $$ $$(((Air(Carrier(( 31.92((( (((((Extreme(Weather( 4.01((( ((((((Security()( 0.13((( (((((Total(( 100.00(

    Delay(Types:((Security$delay$$not$included$in$$analysis)$

    Data(Preprocessing(

    The$ Global$ Morans$ I$ test$was$ significant$ at$ or$ above$the$ 95%$ confidence$ level$for$ all$ delay$ types$ and$months$except$for$Extreme$Weather$ for$ September,$October,$and$$November.$$$

    Extreme$Weather$Delay$

    Results(Aircra9$Arriving$Late$Delay$Global$Morans$I$$

    Table$1.$Reported$delay$types$breakdown$(2012)$

    Table$2.$Global$Morans$I$results$$for$Extreme$Weather$Delays$

    Figure$1.$Airports$points$dataset$with$buffer$size$corresponding$to$#$of$monthly$arriving$flights$$

    Note:$Only$selected$result$maps$for$local$indicators$of$spa2al$autocorrela2on$are$shown.$$Overarching$trends$from$all$months$results$are$summarized$in$the$bulleted$list$.$$C/O(=$Cluster/Outlier$Analysis$HS(=$Hot$Spot$Analysis$

    References$1.$BTS:$2014.$$Airline$OnUTime$Sta2s2cs$and$Delay$Causes$$2.$MITRE$Corpora2on$Center$for$Advanced$Avia2on$System$Development:$$2007.$$Capacity$needs$in$the$Na2onal$Airspace$System$2007$$2025:$$An$analysis$of$airports$and$metropolitan$area$demand$and$opera2onal$capacity$in$the$future.$$1$$45.$$$

    Conclusions(( Sta2s2cally$significant$regional$trends$were$found$to$exist$for$flight$

    delay$type$percentages$over$the$course$of$one$year,$2012.$ In$par2cular,$clusters$of$high$percentages$of$Na2onal$Avia2on$

    System$(NAS)$delay$were$found$to$coincide$with$areas$with$clima2c$trends$known$for$producing$unfavorable$weather$condi2ons$for$flying.$! Snowy$condi2ons$in$winter$months,$thunderstorms$during$spring$and$

    summer$months$! $Result$makes$sense$given$that$weather$was$the$cause$of$nearly$70$

    percent$of$NAS$delays$in$2012$(Table$1)$ Less$clear$are$the$specific$reasons$for$regional$air$carrier$and$aircra9$

    arriving$late$trends,$however$inverse$percentages$of$other$delay$types$and$the$effect$of$2me$zones$and$flight$schedules$are$speculated$to$play$a$role.$

    Spa2allyUdependent$trends$with$regard$to$flight$delays$are$important$in$that$they$are$necessary$to$consider$before$drawing$conclusions$about$delay$at$specific$airports$or$months$of$the$year.$

    $Ongoing(and(Future(Work(

    My$ ongoing$ research$ involves$ the$ study$ of$ changes$ in$ nonUweatherUcaused$ flight$ delays$ in$ airline$ networks$ surrounding$ airline$ merger$events.$ $Given$that$ this$ type$of$analysis$ involves$specific$airports$and$series$of$months$as$the$2me$periods,$underlying$regional$and$seasonal$$$$

    trends$ are$ factors$ that$ must$ be$ taken$ into$considera2on.$ $ To$ do$ this,$ similar$ spa2al$sta2s2cal$analysis$will$be$performed,$but$with$weather$ vs.$ nonUweather$ delay$ as$ the$aAributes$of$focus.$$$

    Figure$3$$

    Figure$4$$

    C/O$$April$$$

    Figure$2$

    Figure$5$$

    Figure$6$$ Figure$7$$

Recommended

View more >