Big Data for Social Good: The developers' journey
Watch this live chat with the teams and participants from the #Hadoop4Good challenge to learn why they chose their projects, hear some of their challenges and delve deeper into the story behind their applications.
0:00 do 0:05 companies are love yeah and we have been 0:08 having lotsa fun talking through and learning about the inner such a good 0:12 challenge 0:12 and we're very exciting point right now which is everybody has shared your 0:17 submission doesn't March 1st and we will be announcing the winners 0:20 on the 15th April so over the course the next two weeks will hope this session 0:23 will have another session like this next Wednesday 0:26 and then all in one they will have to make the announcement what 0:29 was what want competition and we are excited to be able to share that but 0:33 didn't been so far 0:34 let's spend a couple of minutes just talking through each all 0:38 or some of the innovations this text books that are joining us today 0:41 do with that what's your turn it over to bed gonna talk to us about his away 0:45 since that mission 0:46 and what all was motivation and inspiration from the findings 0:50 and so we couldn't you then okay so 0:54 over here tender chicago's Open Government Act makes 0:58 which is a group where technologists and civic-minded 1:01 it together it how it you see also: 1:05 that's they're all costs an area that request there s 1:09 she and %uh she does see 1:12 regularly causes the numbers a problem that 1:15 the commercial developments that are say 1:19 is trying to you actually intriguing what's ours is 1:22 neighborhoods meet their services and it looks as hell 1:27 how to our up may have more sectors 1:31 it's worth the signage is he database 1:35 it is a the hell open government portal 1:38 up this and so that got me thinking I 1:42 party here and we talk about really solution around that 1:46 an asshole looks Oasis or 1:49 so what we did look at all this is likes this is 1:53 city is she going back to 2006 and did eighty 1:57 geospatial ass census tracts and really we're trying to do this 2:01 like I'll across a over it this is 2:04 types where r as urs 'em 2:08 where what services is this supports the Syrian 2:12 up 2:12 ecstasy and then as they were clear around the began to see it was a 2:16 onset oppose a critical business so 2:19 the only one on the site service on the route population 2:23 and so those players bien sur 2:27 date more than that say issue 2:30 those and I'll I should ask followup questions but keep going about 2:36 going to do next which is up 2:39 I have context s 2:42 groups that do SAS social enterprises 2:46 and where he worked with them to see if you use 2:49 worse Israel this is much easier 2:52 but this is going star stretch 12 2:56 make sure that as this is I that it's a 3:00 and sorry deducted with this is a test 3:03 leaks are hits a say Chicago's 3:07 planning sets 3:10 as a stitch that great 3:14 so what did you can I just now four-mile development perspective 3:18 what did you think being able to leverage would you allow did you hear 3:22 that you may be couldn't have done otherwise 3:24 so yearly be able to 3:27 for sale %uh Clara schools the hat was very helpful 3:31 forehead team we all works after hours and on weekends 3:35 and so we %uh where the usual I'll the tools are you sure 3:39 the office clarissa as a group you're doing this 3:43 and it also this or throwing ideas 3:46 trying out seeing where it goes I'll and then just for 3:50 on a to a dataset 3:53 in a way that to we would have though English ace 3:56 well 4:00 well that's good I think sometimes the idea how the fact that you can to try 4:04 things out and do things a little bit more ad hoc is something we will people 4:07 can do with the new 4:08 so that the experience you have also that's that certain goodness mmm 4:12 so are there any question their books purchased many in the group hezballah 4:16 this point but are there any question before we move on to our next door 4:24 all right well we really appreciate 4:26 when your contribution we look well worded to sharing our 4:30 even more those details with boat as you guys do more work and I certainly 4:34 certainly continue to share the community here so 4:37 okay for the next presenter I'll why don't we go 4:41 to thought not when you are you can use yes 4:44 we can do. yes okay cool so I can you tell us a little bit about where are the 4:49 world s yes arm hi everybody 4:53 on who does its our locations that lack access to healthy food options 4:57 and then I'll most a lot researchers have 5:01 try to link this to help issues in those areas some 5:05 a lotta data on this issue has been one dimensional 5:08 particularly at a distance to be grocery stores are examined so what I tried to 5:12 do. 5:12 add more facets up the issue onto my map 5:16 stew basically be realized the most the tasaday 5:20 data um regarding the issue and I hope that that's useful to like 5:24 health care advocates say no public health officials and all that stuff so 5:30 that's really good 5:31 well can you tell us a little bit more about them you're finding so what did 5:34 you 5:34 what did you find out how to use the internet to find 5:38 or yep sure arm so 5:41 this was motivated because what I work in LA I felt like I was in a food other 5:45 I'll and I thought some metrics on with the US government 5:48 said that I didn't work in a few dozen so I wanted to 5:51 you know prove it wrong 5:54 up so what if I'm not was that lot on the southern regions a ballet 5:58 are in particular impoverished areas 6:02 and the distance to the grocery stores are 6:05 actually that bad but the its the density of 6:09 other food options such as fast food restaurants not set that make 6:12 um going to grocery store making you know to be less appealing I'll 6:16 and that's what down on a dip was really helpful just because 6:19 there are a lotta on readily available datasets are 6:23 grocery stores and all that stuff so what up to do was call for that data and 6:27 because I'm calling for at the data comes in all types a format and 6:30 is all tied to weird strange issues and all that stuff so 6:34 I'm I duke was really useful for 6:36 munching the data you know this cleaning it up making sure all the college 6:39 have the right stuff in I was able to join tables make sure 6:42 I'm the data all talk to each other so that I could present and one 6:47 nice format now that's good 6:52 so I do best accelerated that process for you complex you were able to sleep a 6:56 little update is that you wouldn't have otherwise been able to consider because 6:59 you had 6:59 the crop you want call call a place to get there may be appropriate in a 7:04 different type to date at different structures 7:06 definitely owe so much to hope for that okay great 7:12 saw it interesting so well what you end up finding out is that it wasn't at the 7:16 location 7:17 their ability to get to a grocery store was really a problem right 7:21 right right its its the lock-up options arm 7:24 back come from um well one example I have 7:28 is arm cock start 7:31 out Long Beach arm without area down in town 7:35 Los Angeles arm for example that area 7:38 has a very stratified on um 7:41 divers income group there's 7:44 it a lot of people in the impoverished side in this some on this a good 7:48 population 7:48 a I'll very well of folks are 7:52 so in a particular zip code there were 7:55 only one grocer there's only one grocery store I'll 7:58 and on Yelp that the to does a two dollar sign grocery store 8:02 I'll my dataset contains I'm I'm you'll on 8:06 reporter the dollar sign ratings the grocery stores 8:09 and just to give you like a a good metric 8:13 up hope which would be a three dollar sign um 8:16 kinda grocery store and thats you know were Whole Foods is known for being a 8:19 little bit upscale 8:21 so wherein as the code I'm has a 8:24 grocery store doesn't mean that everyone has access to it I'm could be it 8:27 a grocery store that's you know much more I'm expensive or something it 8:31 serves 8:32 its targeted towards one sector up the income group 8:35 so arm I noticed that not it's not necessary if there's a grocery store not 8:40 but this a grocery store 8:42 if is it affordable to the residence in that zip code something like that so 8:47 the question you commit to Russia I think it's gonna cost anyway but how do 8:50 you see this app being used in the real world you know how could you see this 8:53 really evolving into hopes 8:55 yeah I think that consumers and residents in the city of a million other 8:58 cities could take advantage of sure 9:00 on a great question Tasha um I know that um 9:04 recently um there'd been some 9:07 on stock do it done in City Hall where 9:10 the people are trying to pass rules where I'm 9:14 the saying area um the word 9:17 they would not allow AR fast food restaurant to be opened or something and 9:21 they're trying to make 9:22 incentives for grocery stores to be open their 9:25 I'm so any sense I want my apt to be used in that way like I what public 9:30 officials 9:30 to use it to identify places where they should you know 9:34 introduced some sort and sent the producer to be opened their or 9:38 just call because a new fast food restaurants being open their 9:42 on so basically just something to motivate some political 9:46 on development and laws and 9:49 all that stuff right just thought you know providing better access to better 9:53 boot everybody exactly exactly hopefully somebody will 9:57 use it to be there 00 identify alright this is this place is at risk 10:01 and then using the app will be like all-white is there a large 10:04 impoverished population is just a distance issue 10:08 arts something like that what's really interesting you can see how many 10:14 know a lot of metropolitan areas this could be something that's really 10:17 applicable and useful hopes to be all 10:19 to look at so we appreciate you doing that research and look forward to seeing 10:22 wells 10:23 lies ahead thank you okay great 10:27 are things that so I hold my hand right you going make sure you have any other 10:32 questions were yeah yeah I think Jason 10:39 pet alluded to what you're suggesting so would be interesting to see if 10:42 will desert along with poverty areas are lowering heart problems or something 10:46 and what are those correlations that it seemed like that 10:50 that's also what you're trying to get almost like this point if that still new 10:54 that's already indicator what areas are these guys 10:57 you must like most prepared watchin alright 11:04 let's see I think that's good I will not thanks very much we appreciate it and we 11:10 look forward to hearing more about your food desert 11:12 discoveries thanks very much for your participation 11:15 thank you alright cool so again anybody ask questions we will take questions at 11:20 the end as well so we'll create hello 11:22 I'm at any point in the queue name my every move over to Matt Matt I see that 11:26 you're on a 11:27 I'm you that's a great thing and you have 11:30 watching a loose Brett at saw as I'm here inside and I we can't wait to hear 11:34 about that little 11:36 obviously this year was crazy Sol Trujillo welcome all 11:40 and I've included arm someone to the developers that worked on this project 11:44 with me 11:45 out regale are so he can kind of get a little bit more 11:48 technical insight as far as what this project was and how we went about it 11:52 you know first others read motivation so 11:56 up there was a big problem tens of thousands people die every year 12:00 grade CC and there's really no 12:03 I did I guess consumer endpoint with this data 12:07 it's really messy right now and it's it on 12:10 the all Assoc we our goal here 12:13 in watching the spread was to bring this down to a county-level on 12:17 so that people to get more granular insight into where the flu is 12:20 up and really it was it was a look towards 12:24 possibly step extending this to other disease states 12:28 and on to get experts involved 12:31 eventually but 12:36 so really kind of how we how we develop this 12:41 on project through a series of kinda brainstorming exercises and and part of 12:45 our team worker 12:46 by students from Purdue University on 12:50 me we can be ingested a lot different data 12:54 based around the flu armed CDC down a 12:57 on a train air transportation down to ground transportation data population 13:02 density 13:02 I N and social that as well are from Twitter 13:06 to kinda 13:07 those by get that granular look at where the flu currently is 13:11 but then running it through gonna be regression model on 13:15 to figure out where it will be over the next five days 13:19 so 13:23 armenian Font perspective can you tell us a little bit more just pull up here 13:27 that might be new or two what you guys actually I'm what did you learn 13:31 and what did you really find out as you start to sense at this present for that 13:35 period time 13:36 yet so we spent most of our time 13:40 a as opposed to developing the app we actually spent most part I'm trying to 13:44 figure out what was important in this model 13:47 on and one of the hardest things about this this project was there are 13:52 Armenian the main correlation we thought would exist 13:55 and the main important factor which was temperature that we thought would be 13:58 important in predicting the flu actually was not important 14:01 and it wasn't something that we ended up including in our final model 14:05 on soaps we went on a lot of different has 14:08 population density is a big factor in how the flu spreads obviously 14:13 um one's looking index or how the slough the 14:16 physical mechanism I'll spreads which is it somebody costs 14:21 or sneezes with any a six-week radius of another person 14:24 %uh that gives insight into why that's true 14:27 also travel data and talking about 14:31 specifically round trip today being important 14:34 and that's how how counties in Iraq and other solution would spread between 14:39 county's 14:40 and in the final model om which and it was really interesting us the 14:44 these factors became the most important well that's really interesting still 14:49 well-protected perspective as you if you were to take this forward 14:52 you think you have some other right variables in play it would actually 14:55 allow you to settle 14:57 here the variables that are actually matter when you think was gonna spread 15:01 and you can make britain-based 15:02 former well thanks now 15:05 also but I think some further research 15:09 into you can have the variations on the flu and how they interact 15:12 on kinda geographical level would-be arm 15:16 kinda the driving force in in the advancement of this model 15:22 okay and idk I just 15:24 his guy Edgar all I 15:28 yeah as it stands now are our model is admittedly simple as we learned more 15:31 about the flu and about the fact that there are several strains 15:34 strains the clue and every year a the 15:37 it's a completely new strand that's coming out om 15:41 and and and a lot of other facts we had some people from Purdue that were able 15:45 to inform us on 15:46 on some these facts were looks as we learned about this week and realize that 15:49 we were in a bit over our heads in terms ok getting a really good prediction 15:53 model for the flu 15:54 %uh what we produced right now will 15:58 in on some level predict the clue and that's that's 16:01 true so when the flu season starts it'll be able to show where it's going to hit 16:05 most 16:06 and a what should be happening who 16:09 over the next five days our ambitions are to bring in 16:14 some more qualified individuals to help us build out really good 16:17 model with this as a first step 16:20 so that's that's so that was my next question in turns out 16:23 your plans for taking this research forward and although obviously it was 16:27 for the competition 16:27 whole scene is putting out as you get ready for having this information and 16:32 net while gmail will connect your cynicism already doesn't look or or 16:36 where you know your in and we both worked for 16:40 a big data consultancy and then we're up as we as we mentioned 16:45 involved with a bunch of students from Purdue University and one of our primary 16:49 focus is 16:50 at at Percy oh the organization we work in is 16:53 in I income medical prediction and in dealing with medical organizations 16:58 so arm connecting to those kinda resources that we have on a client bases 17:04 will provide further insight as far as how we 17:07 how we properly to in this model in 17:10 in provide further insight to to the end users 17:13 on locally we've seen by a 17:17 a fairly a pleasant on 17:20 kinda reception on this model by 17:23 we were on kinda the local news in and they actually had it on a big TV behind 17:27 him 17:28 I talking about the model is as it was kind of like the weather 17:32 map non so I think there's 17:35 I think there's a significant number of people that are interested 17:38 in this research and and we can can it continue the conversation moving forward 17:43 okay great yeah we saw a news clip that was so totally great to see it was 17:48 welcome news to sell 17:49 that dole will miss you all the best as you continue that work in research and 17:53 look forward to hearing more about it 17:55 right thanks opened or what right any questions feel free to chat a man 17:59 when we can take them here at the end the call alright so why don't we move 18:04 over to 18:05 David I David do you wanna tell us 18:08 you know you should be on me screw your there you go 18:12 you want to hold water Brady Bunch work what million 18:16 be found lol will be great okay sure 18:19 so Brady Bunch Tom gonna be really catchy am 18:23 year some memories on the show as well but basically 18:27 to focus our analysis was renewed that 18:31 his focus on New York City renewed that classes the register buses 18:36 was decreasing and from services that we've seen most of it was due to long 18:41 wait times 18:42 so we 2010 their drivers a long wait times was bus punching 18:46 and that's what we wanted to study but to kind of to gonna take a step back 18:50 and to kind of go back 18:53 into demotivational wiry you want to look at this 18:57 this problem was that we notice in New York City was that the 19:02 the head there was a transportation 19:05 issue happening so there's a lot of people in New York City 19:08 mo said and did not oncars over 50 percent in our own cars 19:14 and also that rely on public transportation simeon motor 19:17 transportation in New York City's subways 19:20 and more and more each year more and more people 19:24 taking subway is and it's happening so fast that 19:27 the reader may increase in ridership was some ways is outpacing the population 19:31 growth in New York City 19:32 so business ever Impex really 19:35 being established date1 is that commute is a scene like over crowding on the 19:39 subways 19:40 you know it's it's so packed to the brim inspected almost a digital platform 19:44 and you're thinking okay like a few years from now has a population 19:48 increases 19:49 you know this can be capacity issue with the subways so we thought was they 19:54 okay we probably rely on alternative 19:57 motor transportations and what we saw was that those opportunities 20:02 and buses I your bass is the networks are pretty extensively covers all the 20:07 five boroughs in New York City 20:08 however to write it the ridership has been decreasing sweet 20:12 Syria seeing that it's being and the utilize 20:15 and where we when we 20:18 dug a little bit deeper we tried to look into it you were people saying oh well 20:23 passes 20:23 the main thing was that the buses were unreliable 20:27 and peeling the onion a little bit deeper 20:30 Reese we saw a lot of people saying that you know is mostly due to long wait 20:34 times 20:34 I'm so with a lot quicker buses yet wait a long time 20:38 so what are the drivers every thought and what are the common issues have any 20:42 bus transportation is I'm buses coming in punching 20:46 passes Cummins coming in bunches the two buses arriving at the same time 20:51 so that's when announces focus on Lake 20:54 if you want to do whatever we times combat 20:58 don't let em in New York City where would be to wait times that I would 21:01 expect 21:02 taking a bus how often does bus benching happen 21:06 and wind as when best bunching happens 21:09 how does that impact wait time so that was kinda what was wrong 21:13 om thought about how 21:16 announces the products or even a bit long winded but coffee that got 21:20 that's the question know that's good so in your case 21:24 so obviously reading transportation dinner hosted in your what other pieces 21:28 of information that will let you go 21:29 allow you to be you bring your and use as your 21:33 I'll you know baseline for driving us or insight and analysis are all times in 21:38 what could be done with want to help people get their you want but the storm 21:41 China so worried we mostly focused on transportation dear 21:46 just because com 21:49 we were measuring did because it was pretty extensive like the way to MTA 21:54 is checking the bus as they caught two passes in the five boroughs are being 21:58 tracked 21:58 Micek every 30 seconds its route idea I'm so 22:02 show try now says is pretty focused like he was 22:05 measuring we time so the buses I am 22:08 but in that's in in itself is pretty complex 22:11 because we needed to estimate when the buses would arrive at each stop 22:16 the only information that we got from the data itself 22:20 was I'm where the buses were 22:23 each interval each 30-second though and how close they were to each bust-up 22:27 but not giving II direct information when they actually write a bust-up 22:33 so we had to do was will use who was that this is the message Saturday this 22:36 is a massive data set 22:38 because every 30 seconds for the buses combo what how do you 22:42 on Alex Ryan Dube was he worked to help us with was 22:45 newly created this big secret tables and I am 22:50 you're using seizing how only jurisdiction everything we able to 22:54 figure out what you when buses 22:56 on estimate when the buses were able to reach to bus stops 22:59 com so you transportation it we only use transportation data but it's pretty 23:03 massive 23:04 but I do but he will work to give us the platform 23:07 order to SB were two I'll we find the data 23:12 lot quicker okay I'll writes a similar with Bennett also talked about 23:16 think it was Ben beginning but he was able to sort you know a rate on 23:19 something that 23:20 possible questions that he was asking Arsenal salts or rated 23:24 well that's great okay also hold on 23:27 I did we really go to keep their any other questions you want to ask before 23:31 we get to the 23:32 next presenter and I think we're going to ask you the same question that we've 23:39 been kinda going thru 23:40 with everybody so how do you see this going well was network that you intend 23:44 to 23:44 to take ordered and see how it might actually apply 23:48 my way that happen ok scanned using a city or how do you see it 23:51 14 someone there i think is a few potential users but I think the 23:55 immediate potential uses for 23:58 anti-aids 28 itself from a management perspective 24:02 because what we saw was interesting was that we pick the 24:05 is this round in Brooklyn he found out that 24:08 I am you know best bench was pretty prevail in late one out of five buses 24:13 came in in and bunches and what would happen is that 24:17 every time a bus would come in bunches it would double to wait time 24:20 so I can imagine myself as a commuter you know it's you know waiting a long 24:24 time and I see two buses come at the same time 24:27 it could be I'll it could get a new nose a little bit 24:30 Creek but where we see it happening is that you know this is 24:34 real time information every 30 seconds this is being produced 24:37 and your you can basically 24:40 see ahead of time when buses are studying a bunch together 24:44 seeing kinda managed at headway so mama control their 24:48 homer manager Nancy a I can see went to Best deciding to get closer to 24:54 together it and I can say no to the bus travel to say he know maybe 24:58 my once stood on a little bit if that give that bass ahead 25:02 a bit more headway in that way they can kinda distribute 25:06 the kinda medicine headway 25:09 the second use cases in more strategically 25:13 they can start freaking out the key which routes have more bass budgeting 25:17 enriched bus stops impact bass punching you know it could be major intersections 25:22 where does 25:22 have slower traffic are more trash I'm on traffic going on 25:27 so then they can start break and that route into %um 25:31 I think the big into select passos's so 25:34 consider stopping at every stop you can stop at Limited stops you stop and mark 25:38 the major bus stops 25:40 net weight is less home delays on the way 25:43 so does this mean immediate impact so we can see straight away okay great let 25:49 super interesting and he had absolutely has applicable Indian 25:52 and your hopefully drives efficiency should you take it forward 25:56 with the MTA New York and other cities so also 25:59 didn't thank you so much for that update and I we really appreciate the 26:02 contributions that to the challenge 26:04 thank you hurried there any other questions but did you guys 26:08 at Chatham and will take maybe 26:09 and we do have another question for the blue balled will do that at the end 26:12 why don't we go over to Tyler on on Tyler 26:16 I there you go and Tyler is going to Pauls 26:20 or husband sure all happen research what you did there 26:24 why would you want to hear about its already at all okay thanks good to see 26:28 you again 26:29 a white walls so 26:32 with with future per week and we started with a simple idea 26:35 you know everybody in the team we go to you go to restaurants 26:38 it's really easy to see what's going on in the dining room in going out there is 26:42 a letter 26:42 view this you know there's maybe TripAdvisor but 26:46 it's really hard to know what's going on in the kitchen a 26:50 and so you know starting with that premise for the well you know why is it 26:54 so difficult why 26:55 there's so much information and it out there why I is that really a problem 26:59 and then so use in boston's Open Data Portal 27:02 we're able to access that data bring it into 27:06 you know IBM clinics in to do and 27:09 we're really fortunate for the Boston tote because they have this one call on 27:12 this one that you called parity 27:14 and that allowed us to link up all like you know five different datasets and 27:18 really give us this kinda hopin' sure 27:20 %uh what's going on you know in the kitchen at these restaurants 27:24 and then a so we you know in just in use in pink sheets: 27:28 a which is be release easy cut down budget I'm 27:32 and then in terms of and I just jamming on the data and figured out what was 27:36 going on without actually are 27:37 exit both be really helpful no you people the team in the sequel background 27:42 that kinda 27:43 you know here's my vacations here's what's going on record as saying 27:46 and then that allowed us to get to can avail just a minimal number columns that 27:52 we could use for our future for application 27:55 %uh and so once we had that you know minimal at you it's a 27:58 if you up then we were able to generate these grades 28:02 for each restaurant based on their 28:05 you know five-year food inspection history I 28:09 and then kinda but I'm too nice visualization based on something that we 28:12 thought 28:13 most people could relate to so since most people had a report card 28:17 that's very relatable they know how that feels 28:19 let's give this restaurant the report card and then 28:22 you know I showed my mom she likes the a so 28:25 a yeah is a 28:29 let me answer question also legal 28:32 well and it's good it's nice to know that you had yeltsin's itself bringing 28:36 what we think will be cheats for you any information you're looking 28:40 were and I think the report card thing not exactly a very small minority 28:44 all what you might get right place there my understanding 28:47 or your Google maybe one somewhere 28:50 so how do you see that going our 28:54 so he is this something that you you plan on taking power is it something 28:57 that might use in Boston 28:59 what what do you think yes so I mean we've actually we've already had some 29:02 users initially we've already been able to 29:05 are get some feedback we've water in Starbucks in just ask people what they 29:09 saw 29:09 are and a so the city bus into spending 29:14 so you know excommunicated we have a meeting 29:18 on Friday morning with the Commissioner who is overseeing these like 29:21 inspectional services 29:23 and a you know he's really kinda put himself out there is a resource for this 29:27 team to 29:28 take a look at our rupert and kinda 29:31 you know is this the best way to greatness restaurants because you know 29:34 the serviceable 29:35 people report cards and you know when designing the end up we really wanted 29:40 to we note we know our consumer smirk and so we want to 29:44 show this is a great which isn't here yet but we also want them to be able to 29:47 go to individual level data 29:49 economic decisions for themselves and so speaking with miss your own 29:52 on Friday will definitely help a and it's a fantastic he's made himself 29:57 available 29:57 in the same a but also you know kinda balancing that with 30:01 that's where consumer and change to empower that consumers so they can make 30:05 good decisions for themselves 30:07 okay great well that's really interesting and certainly something you 30:11 can see what the water or click really sell 30:14 that saw that interestingly with you the best what the most conversations you 30:17 have to keep us posted 30:18 yeah if I'm officer thing everybody wants nor all traveling around the 30:22 router and use it easy one 30:23 holtz's this but i wanna go or don't they are while holding a reason why 30:27 yet so there's been success and I believe New York and San Francisco so 30:31 hopefully Boston sex 30:32 what would be a very good thing your 30:36 okay cool hang on Tyler where you go much do you have questions for you and 30:41 again will take questions at the end for folks that have them 30:44 well what's year more 30:47 won't who are or cool and 30:50 questions but again if you have any others apparently chatty man 30:54 and then let's go on to our last up presenter 30:58 nets and morning and we were already yeltsin you thank you Tyler 31:03 and let fault let me talk about the city bike 31:06 happen but what you're looking to discern what you learned yeah 31:10 all hi Ruth are the one and thanks for going to support him to get this angle 31:15 to handle 31:17 lot of the month fling that on with the phone calls handled 31:20 so fine and I could make it to the Hankel on 31:23 thank you very much for providing this so IBM blue mix black homeless it 31:28 basically or not does too 31:29 get some hands-on onto the technology I'm 31:32 I I work well in a big dick a project but 31:36 I have primary youths cascading to crunch the paper 31:39 I'm big a budget be quarter or not I don't bet on it this is 31:43 I'm be used the word is clouded artist which one of her top so we don't you 31:47 woods 31:47 IBM mom no Globex platform basically for 31:51 process in the paper so disco all dis janice was a colleague will give me an 31:57 opportunity to not who basically 31:59 good other hand sewn onto this platform I and see how the big suits and big sick 32:04 will you know could be utilized 32:06 in phnom efficient manner basically process the paper 32:09 so to start with I mean I've been told they would be posted start 32:13 will you have provided on your talents Post website and 32:16 I feel do no more connected with those Superbike paper because 32:20 as somebody is know who just mentioned about bill 32:24 bus Brady butch a bus both application 32:28 so I think 32:30 because partition is the pero no like in the next few years 32:34 that is going to be took the basically because the pollution levels and online 32:38 but not to all the other things right so 32:41 that's the reason I mean I was at the pit was the city bike decal 32:45 I N basically I started processing this a divided by using big shoots and big 32:50 situation 32:50 initially and as as a band along Lake you're not looking good debate ok from 32:56 different though 32:57 hope fight in different situations I started but we don't look alike and blue 33:01 not cleared 33:02 abuse scandal with your licence with you with your latest on Lake 33:05 when I can compare it with the paid their deposit and after the 33:09 listening to some of the conversation in this angle no i think im like we want to 33:14 let people probably 33:15 the busted up there david was talking about so 33:19 if we can you know basically interpose these different dosage to get up early 33:24 we can have some big difference 33:31 what's really good no study break site you talked about lol 33:34 Citibank it being that the data all you know kinda resonated with you 33:38 what is it something you know you had taken the entire 33:42 up biking in your city is it something is it me 33:45 is sort of an up-and-coming motor transportation where you are 33:49 I would ever get ample me of the 33:52 musical program and people have taken to that program like crazy 33:55 you know what and 8th year people like to go and all 33:59 and use the Website I think because we don't have as much of an infrastructure 34:02 warned Oracle subways and buses is just not what your picture monolayer 34:06 worker supports like or I think that bites but I'll 34:09 reason washington where you are weird but kinda play and 34:12 in help books make their choices about transportation Yamin 34:17 I'll I'm deciding in Columbus Ohio and it is same as 34:22 both you know Denver because we don't talk a lot of public transportation 34:25 you know but they do have a bike in program called the school go 34:28 I think this column was cool bikes or something but I don't think it is as 34:32 extensive as in all city bikes platform 34:34 I and probably well you know in that case basically though 34:39 this column was biking program i think is only limited to book 34:43 always to campus or students at the campus and probably in the downtown 34:47 no I'm other distances to do is to have to travel to the car 34:50 so but basically I mean why I called president could reduce the 34:55 there does it is I got I was gonna come from OC peach 34:58 named with me in India so 35:01 Lake twenty or thirty years but in all it was the 35:04 bicycle safety 35:06 so everybody used to ride bicycle but eventually know it is I think two 35:11 wheelers thought the worker 35:12 leaders I know I'm doctors like basically just take me to do so 35:16 air quality in the CPE I know all the traffic jams and everything 35:20 so maybe you know we could probably 0 35:23 look at bill CP bike does it send learn some things from the 35:27 from you with related to the pollution levels and probably present in my 35:31 sleeping and then it'll 35:32 if let people want to learn from that don't believe somebody can come up and 35:37 then 35:37 what up such a biking but on me and the city its 35:43 okay cool so than do you have plans to go to take this but already there in 35:47 Columbus or other cities now that you've done this research 35:49 or alderman lucky playing out right I mean 35:54 I'm not sure did because you know I'm and portable diners is that I have done 35:58 or using this platform is I think go but we do want basically annoying 36:02 talks about how the genders different genders are using the biking program 36:06 late and basically wall what are the 36:10 basically not the duration of flights that users are willing to pick 36:15 so one of the out order to go to local bar 36:19 up on the plane said that are court of the running few 36:22 big group is basically people are willing to but I do not reportedly 36:26 let's say 20 minutes so if equal Glee 36:29 will put the bike incisions within 20 minutes of each other then probably a No 36:33 upbeat most of people with one won't boot but I work with the word 36:37 using bike 36:42 in okay cool we appreciate you 36:46 you provide health insurance as insights with us I think it's very useful 36:50 and you know it's really interesting obviously we call this the big game for 36:54 social challenge 36:55 and having such a basque have open 37:00 said questioned you guys are able to answer thing is for the social good 37:04 Alta decided to focus on this will open interest and from 37:07 little onto the biking to the transportation 37:11 food for two participants and then we'll hear from everybody else 37:14 right next to it but i think thats really just been 37:18 your YouTube it in and all and good for almost understand because it's 37:22 its consumer all I'll let me just do this to any 37:25 other also questions we can go ahead and take them so please I just 37:29 but yeah I suggest you check in now but or 37:32 I'll there's one more question for Matt in our 37:36 little folks okay and 37:39 that was howdyhi visualization I why 37:43 question i'm looking for. but basically what's the role 37:46 %uh that his visualization might be able to place you take this power 37:50 sure in now I can talk to this 37:54 a little bit arm but we initially kinda went the route 37:58 world I utilizing some kinda 38:01 d3 technologies and stuff too to integrate the 38:05 visualization and visualization is actually kinda one of my 38:09 main focus is rite now in the work that I'm doing 38:12 and so I think utilizing on 38:16 tools like a unity you are like a blender some other 38:20 some other platforms for for gaming technology can provide 38:24 in more rich an interactive visualization 38:27 of were kind of the end user to kinda by 38:31 glean as much inside as they can from and this project was gonna 38:35 her doubt it the visualization aspect bisbee we initially at the 8u 38:39 sat down ok do the initial idea was let's make whether 38:42 with roots and the visualization was I did before content 38:46 and the all this project and really i think is why i like it 38:50 so much is because it really instant and speech to write 38:54 all this DT to see id 38:56 and installations or you know that's good that's I mean certainly something 39:03 very powerful not 39:04 old can relate to. now for you guys and I'm gonna give everybody 39:07 that still on all our prisoners that have are here I think everybody 39:10 all her to spend share our online here 39:14 %uh but for each of these people what was the most surprising thing that you 39:19 learned as part of this process 39:20 so we'll start bankrolling they will do is let's go through everybody's answer 39:24 to that question and will wrap up we have about 10 minutes to go 39:27 I think that those interested in buying because the think their challenges right 39:30 there's no give an answer when you start the process so what was surprising 39:34 what this process I know you talk a little about temperature being 39:37 area well and maybe it wasn't what were there other things sure 39:42 I think on I think someone the main kind of surprises relating to the project 39:47 I itself we're kind of how how the variables interactive what 39:52 what had a strong correlation and what did not have a strong correlation 39:56 but as far as the challenge is a hole goes I think some of them 40:00 strong things that that were surprising or maybe 40:04 just a learning opportunity were really kind of what we 40:07 what we figured out along the way you lie seen on the IBM tools 40:11 I integrating no dread and and the 40:15 IBM on it Lumix analytics suite 40:19 was really kinda be and enriching learning experience for us 40:24 and F lee brought us some tools in our toolbox that we can utilize moving 40:28 forward do anything you want to expand on a 40:31 now tied up the crew 40:33 there is it was a huge challenge one way because we ran into an immigration 40:37 problem 40:37 actually MacBook up between old red the 40:41 IBM analytics for you that you couldn't load more than 40:45 3 kilobytes from between the two 40:48 which was a huge roadblock for a while definitely a problem 40:53 it so this who took us a while to fix it reaction pics 40:56 the that are no bread its I was pretty intense 40:59 we were able to put the Celtics didn't want small 41:03 yeah we were really lucky I'm do was it was probably the 41:07 you know a week before the actual submission date when we found that 41:11 that bug and realized wanna bigger act it was having 41:15 on the final product and luckily I p.m. was very responsive its ok like a day 41:20 for them to which the pics that I made to 41:24 all I B note read okay it 41:28 work just fine so more won't let me use let us know those to benefit 41:33 Michael and we're almost great yeah 41:37 alright cool let's go back to your first presenter bed 41:40 and then what about 30 what was the most surprising 41:43 all happy said what you were able to learn and discover a challenge 41:53 right you know what what we got let's go over to 41:56 at network what they will come back to them alright 42:00 a while 30 um 42:03 our first comment on that data mining dollars 42:07 a big challenge for me I'm I remember reading somewhere though say data 42:11 scientists have to spend around 42:13 eighty percent of their time data mining and I believe that's completely true 42:17 on one aspect of this was although this place the visuals I had to do 42:21 LA had no 42:23 appalling on for all the zip codes but they have a really high fidelity 42:27 um data set for now there's a lot of points 42:30 and then I'll my visualization to blue just couldn't handle all the points 42:35 so I have to do a lot of small stuff that didn't even really relate to the 42:39 data just so much 42:40 presentation at the data output Inc you Jess a.m. 42:44 open source some geographic time information system 42:48 comp reduce the number of points and even know I'm just doing stuff like 42:52 Americans joining datasets on geographical 42:56 locations so like I found a mcdonald's 43:00 and I wanna know what's it called a false and I put in the software and then 43:04 to say 43:05 this the court following the holding on that's great are 43:08 that is a McDonalds in that zip code on 43:11 and this is all done locally on the program but I wish I could have done was 43:15 no pick it apart right some scripts and push all those calculations the 43:19 do you job geometric operations are to head the 43:22 um ever and then in regard to the project itself 43:25 I was reminded that LA is a different city 43:28 on people do food is it's on to differ metric is a row 43:32 metric in a Metro metric um what it means is that usually 43:36 grocery store in South Ardmore further in rural areas just because they're more 43:40 spaced-out 43:41 and then when your new york they're more deaths so through dozens 43:45 have different I'm definitions sometimes but I'm for 43:49 a late a lot of people just visiting LA notices to its a pretty 43:53 big-city its its their sprawl so 43:56 what we notice is that not only is our poverty and all that stuff but 44:00 distances are very unique and LA purses 44:03 other cities like New York on band and in Chicago and so for 44:08 so about something that I i that was unique to the LA 44:11 she does its you know that makes no sense 44:16 alright cool tell what about 30 44:22 ISO I'd say there's probably two parts to it one is the the features that we 44:27 didn't implement the ones that didn't make the cut 44:29 that we were really kinda really wanted to badly but it just 44:32 wasn't quite there and then the other thing is he going to start with all 44:36 these different 44:37 you know occur almost disparate data sets and you have all this information 44:40 and then you need to 44:42 it's good you know said don't think about it and then 44:45 how do you presented an intuitive way a you know 44:48 is this really a lot of information than anyone ever tell the most important 44:52 story 44:52 and so actually walking through that process before even touch the computer 44:57 I'm does you know those a challenge what's most intuitive way to preserve 45:01 all this information and then just the features that didn't make it in 45:05 were so we had a we have this one feature where we want to hook up 45:09 a you know link to: so you have your food grade and then you have a link to 45:12 the opening and see what the people are saying 45:14 and so we're at WebCrawler and like all this stuff and you know 45:19 I saw her and twisted and just as the the you know pretty big effort and it 45:23 just 45:24 at the end of the day we just didn't have time to implement also 45:27 you know hopefully that makes certain future so mean that would be a fabulous 45:31 channel are you guys because I was there or not slow 45:34 yep I'm just getting home value at also they're using L 45:38 Yahoo to take revenge that had a DSL yeah 45:41 so yeah features that didn't make it and then just really kinda 45:45 before touching the computer like you know really thinking about how to 45:48 had a capture intuition with the data so also I'm not hopeful 45:54 now let's go back to bed and then we'll and what they can hear you this time 45:58 money cannot hear your 46:03 marina I'm in total net went into his loss these my connecting them 46:07 okay I'm is there anything else you wanna hear around 46:10 anything you can't really the discover Orange Bowl chance with this process 46:14 yeah I mean for me the most challenging part was to build you 46:19 rejoinder traditions because I not this 46:22 rowly was the first time I was like trying to visualize something ok up the 46:26 paper 46:26 iron so that was one of the most important thing that you know I 46:31 almost talent in NorCal this whole lot 46:34 challenge that's a good thing and are another surprising inside the iPhone 46:39 old phone or up to you know I blocker does some Krauts is basically during the 46:43 winter 46:44 of I'm in the last year's winter 2013 going to fork in the road approximately 46:48 500 kV riots 46:49 taken by the Bob bikers so even though the templates those 46:54 are no sub-zero so people are willing to ride the bike 46:57 though if that I lick special bike group probably be good how some more 47:02 bush remotely including broom been another 47:06 in third but I found is basically little bit early 47:09 good in this number of female drivers compared to the male drivers 47:13 which was expected but it became a breeder 47:17 once you actually plot something in a good look at it and then 47:20 basically gives you a greater impact than just telling the numbers because 47:23 you can see in the by 47:25 but the very small but a 20 per person does Oprah no criminal record 47:29 celebrating the 100 bucks 47:30 cool iron lekin other 30 only in new york city there are so many bike right 47:36 by groups 47:37 which are readily available to people under Dubai 47:40 so but basically I think that is what I know that helps 47:43 in phnom ridding the city bike program are not successful 47:47 working quite well we appreciate that 47:53 and didn't get too wet slot let's go David walkway 47:57 David the question while I what was ya for the developers for those that are 48:01 you that 48:01 udall this hard work what was you know maybe the most challenging her 48:06 surprising 48:07 MTO and a I don't steal 48:11 so I think media air must I can hear you he called me 48:15 hey sorry I had some connection issues but on but I'm back on land if you can 48:19 hear me 48:19 weekend what can your God also you what we can hear you so 48:23 will go ahead and yet the event had a final comment on that on that question 48:27 let's go ahead and do that 48:28 okay a touch briefly I am 48:31 said development so for me I think 48:35 for me it was more street for it I think lost the ball quick was doing and ask 48:41 this but once I had the analysis and 48:43 what I wanted to you 48:45 so the variables are features I want to show are basically 48:49 I E headed in a big secret able and then I was able to connect the truth tableau 48:55 to get some a Divis allies asians out so from you 48:59 mall that was in the nasa states and then for the development 49:02 use all I do a it was easy to connect from there 49:07 for tablet to the big secret tables hope you didn't answer the question 49:11 well that's good that's very helpful and where to go 49:15 three's a charm back home update 49:18 you know you wanna there you got a warm canary 49:22 OK Corral question and then what animal close-up 49:25 okay for me i i the fall asleep Acer 49:29 interesting outcomes that this is that theirs is a technology 49:33 it was created by a for large part is created by marketing firms 49:37 try to end point people the southern states we began its every bite 49:41 use this technology to help identify communities 49:45 don't know they exist so for example people that 49:48 up that for whom some 49:51 only parks or is the only thing they have and what up 49:55 they did you know that as their club this autocorrects or so healthy people 49:59 figure out their community maybe help them to band together 50:02 you interesting things up and there is a 50:06 for us in the the technical side I think be 50:09 week at whatever stage was a so lucky little 2015 50:13 that her so what is this data is available police for dole's 50:17 some framework still in this analysis and to create a Yahoo cases 50:22 so it was a we agree a serious look this together 50:26 and we some people ozzie while know that's really helpful 50:32 and I'll all realize what's at stake here you know met late 50:35 and I was started a couple minutes late but I just wanna thank 50:38 all participants join today your insight 50:42 are incredibly are all we appreciate your work and tying 50:46 and hopefully those who have had the opportunity to attend a gallon 50:49 you not yet to really hear about %uh what you consider it what you are able 50:53 to find out 50:53 so I will just leave even 2 comments here don't have to be sorry 50:57 accorded the answer is yes so we will post those instance their record or 51:00 OSA looking back in a minute but they weren't able to attend your 51:04 they were and then we'll be back again next wednesday at the same time 51:07 her me what's on the other side mirrors so thank you everybody for your time 51:11 today and I will show a great afternoon 51:13 have a great day thanks guys your YouTube 51:17 Press tomorrow do 51:22 do