 Hello, my name is Jose. I'm Liz. I'm Michael. And I'm Juan. And we also have Rachel, but she couldn't be here today. This semester we worked on a project with the Children's Healing Center. We attempted to answer two open-ended questions, which we'll get to in a second. But first, I'd like to tell you a little bit about the center. So the center is a nonprofit organization that has been open since 2015. They provide a space for immunocompromised children and their families. They currently serve about 360 families. Because of the nature of the center, they put a big emphasis on the well-being of the members. And so to accommodate health-related wellness, they ensure that every visitor, when they come in, goes through a process that encompasses basically a wellness checkup. Make sure they're healthy. They're not sick. They essentially leave as much of their personal belongings behind so that they're not bringing in too much into the center that might be potentially contaminated in some way. And also they emphasize the mental and physical well-being so they have different rooms in the center that are designated towards, for example, active play, learning about technology, exploratory play. And so with that being said, Liz will tell you a little bit more about the details of the project that we've been working on. Okay, so when we were approached by the Children's Healing Center, they were currently serving about half as many families as they potentially could be. So our goal was to help the center grow efficiently. And so that was our overall goal. So we kind of broke it down into two smaller questions that we can use, our data to answer them. The first being, we wanted to identify who the center served, so identify the characteristics of the families that benefit the most from the center. And so our second question was to minimize the cost of acquisition, where acquisition is the time and resources it takes for the center to get the families to come into the center. And so we wanted to identify the members that had the lowest cost of acquisition to minimize the amount of time and resources the center needs to get these families in, but also identify the members who will benefit the most. And so the Children's Healing Center gave us several different data sets, but the main three that we utilized in our analysis were the visit data set, the context data set, and the membership data set. And so the context data set contained general information like their gender, their member type being like if they were a parent, a sibling, or a qualified member. Our visit data set contained several different things. So it had, they tracked their every visit for every family. So it tracked the date that they came in, the time that they went out, and the time that they were there in minutes. And our membership data set contained general information about these members. So they contained like zip codes, family size, and household income, but we couldn't really use household income since there was a lot of missing values. And so this is an example of what are the raw data sets that we got. And so as you can see, basically none of it is, it's all very formatted, but you can also do that household income as several different, several missing values as well. And so to identify the members that were a high benefit, we wanted to quantify this definition of high benefit. So from the data that we were given, we were able to quantify it three different ways. So we had three different definitions for high benefit. The first being the number of times the family has visited the center, and that was calculated using the visit data set which contained information about each time the family visited the center. So we just summed the total times that they were there. The second being the amount of time in minutes that the family has spent at the center. So that was calculated using the visit data set as well, which contained the duration of the program in minutes, so we just summed all those. And the last being the total member days of the family. So that was calculated using the membership data where it had the date when they first came in and their last date that they were there. And so we just calculated the total days between those two dates. And so high benefit, we basically, for each definition, we found descriptive statistics and we divided up the groups based on these statistics. And so each definition, we wanted to use everything above the upper outlier to ensure that everything above that were defined as high benefit. But for this example, this is total member days, and we can see that our outlier doesn't contain any data points. So we actually use the median for this one so we can get more people in our definition for total member days. All right, so these demographics that we chose to look at were the condition of the qualified members, such as cancer or in our hospital diseases. We also looked at age ranging from infant to age 26. And as for location, we have the zip codes of the numbers, and we found the drive time and dates that it took for them to get to the center. And we also looked after the sizes. All right, so now that we've had some characteristics for high members, we wanted to compare what these characteristics look like compared to the total population. So we used one of our three definitions, we used the total days and compared it to what the total population of the center. And this first one, we're looking at condition. We can see that the high benefit is almost identical to the total population. So we do see that condition isn't an indicator of whether they're high benefit or not. We found this relatively early. We were kind of surprised by it because we thought it might be. As we can see, since the data is almost identical, we know that it's not a good indicator. So next, we looked at ages. As we can see, the ages in the total population were more like the four and the five and the six. But when we looked at the high benefit, you can see our range was actually five, seven and nine, which we found interesting. We know that it seems for total days at least that five, seven and nine were definitely a better indicator of what's going to be more high benefit, even though there are more kids aged four going to the center. So next, we looked at location. So location, we took the zip codes and developed approximately a draft time. So we found that those families who live closer within 15 minutes usually, we're going to be going to the center a lot more, which kind of makes sense. Because if you live farther away, you're not as likely to be driving to the center or a day or so. So we found that, like we said, 15 minutes is pretty much the only standout as far as compared to the total population in our data. And then we looked at family size. So as we can see, this graph kind of looks normal, which makes sense. It says we know family sizes of four, five or more common with two kids, two parents. So again, this kind of matches up pretty much dynamically, except for family size four, which we did find was actually more high benefit than what our total population says. So we were able to kind of say that these family size reports are benefiting more from the center than what the normal population says they should. So towards the end, we looked at one-time visits. This is these families who are coming into the visit one time, and that's it and never showing back up. So for here, we looked at family size and draft time. We didn't really look at age and condition because, again, that was pretty identical to what the population said. So we were able to kind of ignore them because it didn't really give us anything as far as one-time visits. So we looked at family size, and we were kind of surprised that families as the five were only showing up once. We figured since they have this family of five, they might want to be coming more often, but after talking to our agents, they made a good point that with this family of five, they might have other responsibilities to their other children and whatnot, and might not be able to come in as much. The other one we looked at was location. So we saw that, again, this makes sense that the farther they are, the less likely they are to come back. So these families that live 35 minutes away, that are way more frequent than the population, makes sense because they're probably not going to want to drive as much. And then as for cost of acquisition, we have draft recruit time, which is a lot of time and days between the referral date of the family and the date they joined us after, and we compare it to membership length and date. And as you can see, the side of it is a really great job at recruiting members. So everyone has a pretty low recruit time. And ideally, we would have all the members in the upper left-hand corner, which would indicate that they have a low recruit time and high benefit as well. But we do find that the members who are in that category have similar demographics to the high benefit at this cruise. And to wrap things up, we were able to define characteristics of high benefit members, which we outline as they generally live within a half-hour distance to drive time. They have a family size of four to six, and their ages range between five and ten as far as the members are concerned. When we looked at minimizing the cost of acquisition, as mentioned, the center does a really great job at recruiting members, people that are eligible, families that are eligible to come in, but their retention rate is low. Another thing that we saw as things to consider is that the timeframe that the center has been open is small, and because of that, we have a small DST. And then for suggested future work, we think it would be good to analyze income once more data is available. It would be good to start a survey of their members to see, to attempt to answer some of the questions that are more soft questions in the sense of like, why are people coming? In that sense. And we would like to see a repeat analysis on a bigger market, they have said, and also to investigate retainability. What can they do to help retain their members? And lastly, I'd like to thank our professor and advisor, Dr. David Austin, our liaison, Amanda of our board, and Big Match for making this possible.