Justin Elszasz @ City of Baltimore | Data governance to better serve citizens | Data Science Hangout

Transcript#

This transcript was generated automatically and may contain errors.

Welcome to the first Data Science Hangout of 2023. Hope you're all having a great week. But if we haven't had a chance to meet yet, thanks for joining us. I'm Rachel. It's nice to meet you. This is an open space for the whole data science community to connect and chat about data science leadership, questions you're facing, and getting to learn about what's going on in the world of data across different industries. So each week we feature a different data science leader as my co-host to help lead the discussion and answer any questions from you all. So together we're all dedicated to making this a welcoming environment for everybody.

So I love when we can hear from everyone, no matter your level of experience or area of work. And if you've been here before, you've heard this feel a lot, but there's always three ways you can ask questions, but also provide your perspective. So you can jump in by raising your hand here on Zoom. You could put questions into the Zoom chat. And feel free to just put a little star next to it if you wanted me to read it from the Zoom chat instead. Otherwise, I could call on you to introduce yourself and add some context. But we also have a Slido link where you can ask questions anonymously. And so I see Tyler and Hannah, thank you, both just shared that in the chat. But we share the recordings of each session up to the Posit YouTube, so you can always go back and find helpful resources and see an overview of each week. And we also have a LinkedIn group for the Hangout, too, which I'll share in the chat here in just a second.

One other note I wanted to add here in the beginning, too, is if anybody is hiring right now and wants to share that in the chat, feel free to put the link to the job descriptions there. I was going to share a link in a little bit later for Chase Carpenter, who is one of our featured leaders from the Chicago Cubs. They're hiring an assistant director of database marketing.

For our first Hangout of 2023, I am very excited to have Justin Alsaz here joining us. Justin is Chief Data Officer for the City of Baltimore. And Justin, I love to have you introduce yourself and probably fix where I messed up on pronouncing your last name and share a little bit about your role and maybe something you like to do in your free time, too.

And honestly, for some of them, there is a pretty direct, even though it sounds ostensibly about, you know, internal capacity or being inward focused, there's a pretty direct line and theory of change from if we improve data literacy by building this data academy program, we will improve service delivery to our residents.

Data-driven service delivery in Baltimore

No, I got there in time. No, so I live in Baltimore too, and there's definitely a lot of issues here that I'm concerned about that I feel data could really address. But part of that is making sure that you're able to collect the right data, that you're able to do analysis on it, and that someone listens to that and makes a decision. So I wanted to see if you had any examples of where that has been done or where you see it heading. Like, will 311 be better? Will street sweeping be better? Can you make sure certain neighborhoods aren't underserved?

Maybe a couple of different scales I can offer. So one about, well, this would have been just before the pandemic, so I cannot believe it's been three years. So we identified a pretty big backlog in DPW in terms of cleaning service requests in the city. This was upwards of, I want to say there were 15 or 17,000 service requests. That was a backlog. So these are things that had accumulated. It was right in the middle of the summer.

One was we worked with DPW basically to even identify that as an issue and start discussing how we were going to deal with it. And part of that was actually, we collected a few parameters from them. And this is more of like an engineering approach as opposed to a data science approach, but we built a small kind of heuristic model where you could modify kind of the number of crews that were online and some of the distribution of where the SRs were and that kind of thing to try and estimate how quickly we could drive things down. And so you can manipulate how many crews you have where in the city and you could see whether we might or might not get to clearing out the backlog and when. Because I think we had set a particular goal of getting that eliminated by a certain date.

So one was, and that was a bit of a process of convincing that, yeah, you can ask for their parameters and plug it into your little model, but all they have to do is say, I don't know how this model works or I don't believe your model to like not want to drive their decisions. They're not going to listen to it. So it was a very simplistic model that they could, you know, you took them along each step and say, okay, if you've got this many crews here, that means you could complete how many SRs, you know, that day.

The other half of that one was they used to divide the city up into quadrants and they had hard restrictions on how many crews they would put in each quadrant, which meant if a quadrant, if a crew called off in a quadrant, it impacted that quadrant, but no one else. But the issue was if all the SRs, if all the demand was in another part of the city or in that part of the city, you couldn't pull a crew from elsewhere that had a low workload into that area. So it was, they were placing arbitrary constraints on their deployments. We got them basically to eliminate that way of working, that kind of strategy, because we proved out that you're exacerbating the issue by, you know, arbitrarily placing these constraints on your crews. So that was a really good win. And by the way, I think there was a Sun article at the time about it. The southeast quadrant of the city was getting far better service than the rest of the city. So there was a massive equity issue. So again, just by some simple analysis and showing them and bringing them along, we were able to make, you know, convince them to make some better decisions.

In a lot of the, I mean, there's kind of the diminishing returns, right? Like, so 311 definitely covers a lot of our city services and certainly for something like public works, you know, 311 is meant to be the record of, you know, the source of truth. And so, yeah, we direct residents to 311. That's what we look at internally to determine service levels.

And now that said, we continue to have kind of quality issues around 311, like you raised that, you know, for example, there's still the concern that something gets closed, but the work wasn't performed. That issue, like, it's completely dependent on what SR type you're looking at, because there's 200 service request types and each of them is an entirely different workflow. It's a different crew. It's different people involved. So anyways, all that to say, like, I would say on the whole, we're collecting the right data for some of the bigger areas.

I think there are some smaller areas where we definitely need better data. One that comes to mind is, it's related to illegal dumping, but our transfer stations, you know, where a small hauler truck could go and drop off their loads, you know, maybe they're cleaning up debris from, you know, a construction site or something. We have really long line waits at those drop-off stations. And we've always had the, we've always hypothesized that because there are those long waits and these guys are just trying to make money, that if they see a long wait, they're going to leave and go dump their load somewhere they shouldn't be dumping it. And so that exacerbates illegal dumping issues. The problem is we don't have either a digital licensing or check-in process for the transfer stations. So we can't actually monitor, you know, if you have a small hauler license, how often are you actually checking in at the transfer station? Because if we see that you have an active small hauler license, but you're never checking in to a transfer station, that might be a signal that they're taking their loads elsewhere. But because we don't have any kind of digital tools for doing that check-in process, we aren't able to look at that issue all that closely. So there are definitely pockets where we absolutely need better data. And again, it's always a matter of cities or resource straps, you know, picking, and like you said, there's a spectrum of kind of leadership in terms of being data driven. So it's finding the areas where you've got all the right constraints satisfied to do some good.

Modern tooling and code governance

Sure. It's been fascinating to hear your stories, Justin, and really exciting to hear these perspectives. So I'm just kind of curious as somebody who really appreciates, you know, the tool stack and how we're taking modern approaches to data science, what kind of modern tooling are you trying to implement across your workflows, especially in the, you know, the government sector where, of course, my collaboration mostly with colleagues at the FDA, and I know they tend to have some strict requirements on what they're allowed to use and whatnot. So just kind of curious how you're handling those approaches for like best practice of co-development or releases within your team.

Yeah, it's a joint effort with our IT department because, you know, for the ransomware attack we had, so we had a ransomware attack here three years ago, and a lot of cities have experienced these over the last couple of years. It really ground city government completely to a halt. I couldn't access my email for more than a month. Coming out of that, there's been a pretty strong emphasis not only on cybersecurity, but even just modernizing and professionalizing our IT approach, and so our IT department has grown by leaps and bounds in that regard. It's a joint effort because there's, this is one of those spaces where there's a little bit of overlap between IT and the technology side, but also data analytics and that side of things, and the data governance around it.

So things like, we're just now getting into like what exactly is our code repo policy. Internally, we have Azure, like the IT department's using Azure DevOps now for code and deploying, and so getting our data analysts onto that and getting them to recognize the value and bringing that into Azure is important because, of course, on the data side and a lot of the civic technology side, you also want to build open source tools, and like that's always my preference as well. But a lot of folks who either come into data science from other areas, they aren't always up to speed on kind of software development best practices, and so you do get folks, you know, writing code, maybe code that doesn't need to be or doesn't belong on GitHub and really doesn't actually need to stay internally. So that's something we're actually kind of confronting right now is let's get together a policy where we all agree this is a great use case for open source. This is where you should make something open source, and here are the places where we can't make it open source and need to keep things internal. So there are definitely governance issues there. That's kind of a live question for us.

No, that's quite interesting to me, and I can sympathize. In fact, I was on a meeting just before this where my colleagues had never heard about GitHub in this upper part of the org, and yeah, there's lots of education that we even have to do in our own organizations about some of these tooling. But yeah, certainly sounds like you're thinking a lot of the great ideas, and I agree. There definitely is a lot of overlap that can occur between what you're proposing from a development, you know, tooling perspective and what IT is trying to serve up, so having that relationship solid up front is always really beneficial.

The other, and the other really big aspect of it for city government or any government really is that we're an organization of organizations. So in some respects, yes, we have central IT who can support all of the other agencies. Up until the last couple of years, a lot of agencies pursued their own IT, you know, whether they weren't getting good support from central IT or weren't moving quickly enough or want to try their own things or wanted to use something different than everyone else was using. So there's been a lot of, like, so there's been a federated approach very distributed, and so, and that applied to the data folks as well.

There's an argument to be made that there are things we need to centralize, and what were constraints around? The concern is always, well, we don't want to slow down an agency. If they've got someone who knows how to use data over there, we want them to, and the agency knows what they want to do with it. We don't want to be a bottleneck, and we don't want to constrain them. We want to get them to use that data how they need to. So there's kind of a balancing act there where, like, yes, we need to be a little bit more constrained and a little bit more centralized with some things, but not at the cost of innovation, agency responsiveness, you know, real problem solving that needs to get done. So it is a challenge when it's not, in some respects, we're one organization, and in other respects, we're 30 organizations.

Community listening and project prioritization

Hey, Justin. I'm Rory. I live in Baltimore. I'm in Canton. A while ago, I don't know if you guys have come across this yet, but on Reddit, there's a thread of someone scraping the water bill page and taking a look at who has outstanding bills, and in general, the gist is that there's, I think it's like $330 million in water bills that haven't been paid, a lot of them actually being, like, business or corporate customers. I was curious, because right now, I work in people analytics, and I'm working on building out a sort of active and passive listening strategy, and so this is what's top of mind for me right now is, how are you guys sort of collecting the needs of the city and deciding what projects the data team can get involved in and really help solve some of those issues?

Yeah, that's a good question. I think every team and I think every office or department or agency or whatever has slightly different ways of doing that. And for, you know, once we actually have a, you know, essential like analytics team as opposed to city stat or some of the other data functions, I think like we'll, some of the criteria that of course come to mind are, you know, having a project where you've got leadership buy-in so that if you dive into the problem and come up with something that they're actually going to be responsive to it and know how to approach it.

As far as listening goes, there's, like I said, it depends on the area, but there are a few different conduits. There's, you know, there's always going to be that issue that the mayor cares about. There's always going to be issues that the city council cares about. The mayor's office of constituent services who kind of deals directly with residents on things that are like maybe not just a straightforward 311 request but are maybe more lengthier in-depth issues. Agencies all have their own community liaisons and PIOs that are out in the community and surfacing issues.

For a data team, you know, like, yeah, and how a data team might choose based on all of these different inputs and communication streams. Yeah, like I said, leadership, data quality is another one. Is this an issue, is the problem at hand that we have data that might help us address the problem and we just need to put some brain power behind it? Or is it, well, when we actually dig into it, we don't have data at all or we don't have the right data or good enough data. And so now what we actually need is to invest in a new platform or create a new 311 service request type or put these people in contact with constituent services. I don't think there's any one cohesive strategy, like, community engagement is just messy.

And I try to regularly do kind of listening tours with the agencies themselves. So whether it's talking to agency heads or, you know, deputy director of this bureau or even the data analysts themselves to surface issues. Yeah, there's no shortage of, there's no shortage of issues to address or even brain power or good ideas. A lot of it is there's just limited ability to execute. So it's the real battle is prioritizing and trying to keep focused on something for long enough.

Yeah. I also think that, like, the, you know, the hundred thousand foot level, you know, for Baltimore, like, the priority areas, of course, are public safety and city cleanliness. And so the issue is that, like, solutions to both of those can be completely based on neighborhood, you know, within the city. It's very hyperlocal, both of those issues. And so for issues like that, your qualitative information, the voice of the residents is just as important, if not more important, to make sure you've got a sustainable solution to address those problems. The data that we have centrally to maybe analyze some of these situations may or may not tell us something at the, at a very local level. We may have, so for one data, yeah, we can zoom in on a particular block. But for other issues, we may not have data at that scale. And so you've got to go listen to the residents.

Change management and data governance lessons

Thank you for hosting again. And, um, it's a really interesting talk. I'm always trying to like translate things that people bring into what we're doing at my, um, at my research institute. And, um, yeah, my question was sort of around, how do you, what are some key lessons that you had? Um, because you're trying to do, it sounds like a lot of change, a lot of scary change to a lot of people and resources are limited pretty much everywhere, but you know, specifically in the public sector and, um, I'm in academic research. So, um, I was wondering how do you, um, there are some key lessons as to how to navigate, uh, that sort of change leadership part of things.

Um, one thing I've learned, and I think this was probably a mistake of mine early on in the CDO role was, um, there's, so when it comes to data governance, right? Like it's, there's a variety of different issues you could be tackling. There's different approaches. What, what I had convinced myself of was the need for, let's stand up this broad data governance committee for the city. Every agency will have a representative. We're going to take one issue area at a time, but everyone is, every agency, every representative needs to make some progress on that area. So if it's data quality, you know, we created this data quality guide book or toolkit.

I, I think that was probably the wrong approach, especially in an environment of such limited bandwidth. Um, not only from the agency side, but from my own bandwidth, I think, um, I'm having a lot more success and doubling down on the areas where there is obvious buy-in and people want to change something. Um, trying to get everyone to move an inch in the same direction was a real grind. And I'm not convinced that it has the same impact as finding the one or two people who want to go a couple of miles with you. Um, so that's, I think that's where I'm, I think that's one of the, been one of my big lessons learned over the last two years.

Trying to get everyone to move an inch in the same direction was a real grind. And I'm not convinced that it has the same impact as finding the one or two people who want to go a couple of miles with you.

Um, of course, all of the other typical change management advice applies, right? Like, um, listening and making sure you're actually solving someone else's problem and not just your problem or what you perceive as being the problem.

Even, even when you are the person that has the authority to say, yeah, you have to do this. Um, look, I, I report to the city's, the city administrator, who was, like I said, it's kind of like the chief operating officer. Um, they early on were, were involved in setting some of the expectations around data governance. But agencies are under countless, countless unfunded mandates, right? That like the mayor wants them to do this. These are the things that are legally obligated to do. They've got citizens wanting them to do, go in this direction. Um, all of whom have a right to claim that like they have a voice in what that agency is doing. And I'm just one of those voices.

So yeah, like I, I have, you know, I'm, I'm, I report to the city administrator. I'm the chief data officer that comes with some authority, but you know, that doesn't, that still doesn't mean the thing's going to get done. Um, and not only that, um, if, if I take the approach of, well, I'm in the mayor's office, so you have to listen to me, it's going to backfire and people aren't going to want to work with me. Um, so that card is best kept for as long as possible in your back pocket. Um, because it just, it just burns bridges so quickly. So, um, yeah, the, the idea is really no matter what to find a way to, to ensure folks that what they're doing is in their best interest, um, genuinely, not just like selling them, you know, on something.

Cross-city collaboration networks

Um, well, Justin, I forgot to ask you, do you have a like hard stop at the top of the hour? Timothy, I saw you had a great question on getting buy-in and leading to change. Um, well, Justin, I forgot to ask you, do you have a like hard stop at the top of the hour? Okay. Um, Timothy, I saw you had a great question on getting buy-in and leading to change.

Um, well, Justin, I forgot to ask you, do you have a like hard stop at the top of the hour? Um, well, Justin, I forgot to ask you, do you have a like hard stop at the top of the hour? Um, well, Justin, I forgot to ask you, do you have a like hard stop at the top of the hour?

Um, well, Justin, I forgot to ask you, do you have a like hard stop at the top of the hour? Do you have a few minutes to go over? I'm good. Um, yeah, I don't have a hard stop. I'm good.

Um, yes. Um, that's, I often tell folks, that's one of the best parts of working in government is, um, we don't have to worry about, um, protected information or, or competition. Um, and so there, there are a few different networks, um, a couple that we participate in. So like Bloomberg Philanthropies, um, is pretty big in like the local government space. So Michael Bloomberg's kind of theory of change was that if you really want to make impact and impact people's lives, like local government is the place to be because, you know, at the, at the federal level, yeah, you've got very wide, broad impact, but it is very hard to see like, you know, this trillion dollar budget trickle down to see how it impacts someone's lives. Local government is where that money eventually comes into your programs and projects.

And so, um, the Bloomberg network, you know, there are a couple of different networks that they run. There's the innovation teams. They all, you know, I mean, within the first couple of months of me joining the I-team, we were, we were visiting Los Angeles to meet with the other I-teams, um, and specifically with Los Angeles' I-team because they were working on the same, um, issue as we were going to be working on, which was police recruiting and hiring. Um, there's, they also run something called the City, um, Data Alliance, which is a cohort right now of about 20 or 25 cities. We get technical assistance around a particular track. Um, there's executive leadership training for, for using data.

So that's another network. I'm part of, um, um, it's called the Civic Analytics Network, which is a network of chief data officers, um, from cities around the country. Um, that's run out of, uh, Harvard Kennedy School, um, um, by Professor, uh, Stephen Goldsmith. Um, he was a mayor of Indianapolis. He was a deputy mayor of New York. Um, so yeah, that's, like I said, that there's tons, and, and through that network, like I've, I'll call up the chief data officer from Pittsburgh or Boston and, like, ask them if they've, how they're tackling this problem or what piece of data stack they're using to, to solve different things. So, um, that's, that's one of the best parts of city government is just being able to, like, cold call people from other sides of the country and figure out how they solve something.

That's, that's one of the best parts of city government is just being able to, like, cold call people from other sides of the country and figure out how they solve something.

Thank you so much, Justin. I tried to capture some of those links and share them, but maybe after I can email you for some to share with the recording too. Sure. Um, because I saw a lot of people are sharing different open, uh, data portals in their cities too, which I can collect those all together as well.

Thank you all so much for joining. Thank you, Justin, for sharing your insights and experience really, really appreciate it. This has been a great lively discussion for the first one back of the year. Cool. Thanks for the invite, Rachel. This is great. Thanks for all the questions, everybody.

Justin Elszasz @ City of Baltimore | Data governance to better serve citizens | Data Science Hangout

Transcript#

Justin's background and role

Building data literacy in city government

Data Trails and hyper-local data science networks

Office structure and building an analytics team

Legacy systems and cloud infrastructure

Goal setting and balancing internal vs. external priorities

Data-driven service delivery in Baltimore

Modern tooling and code governance

Community listening and project prioritization

Change management and data governance lessons

Cross-city collaboration networks