Lauren Rogers - Join the (Open-Source) Party
Welcome to "AigoraCast", conversations with industry experts on how new technologies are transforming sensory and consumer science!
Lauren Rogers is a freelance sensory scientist who fell in love with sensory in the early 1990s when she worked for Dalgety, a food ingredients company. Later, she worked for GSK, mainly on beverages such as Ribena and Horlicks. Now Lauren works with several different clients in a wide variety of product categories and also lectures on sensory science at the University of Nottingham. Lauren is particularly interested in the sensory, consumer, and emotional attributes of brands; sensory shelf-life; panel performance; and sensory claim substantiation.
Lauren's Contact Information:
Transcript (Semi-automated, forgive typos!)
John: Lauren, thanks a lot for being on the show today.
Lauren: Thanks for inviting me.
John: Wonderful. I love having guests like you on the show. People who have been sort of through all the walks of life and sensory because you've been in industry, you are a consultant, but you also have been involved with academia. So you've kind of seen the full spectrum and you're 25 years of experience now. Is that correct, Lauren?
Lauren: Yes. This month, I'm 10 years working as a freelance sensory scientist. Thank you for inviting me in November and celebrate my 10 years.
John: I have champagne right here on the air. So, I mean, obviously over the last 25 years when I think back what life was like in 1994. That's when I went off to college. And it's amazing how much the world has changed. When I was in college in 1994, a friend of mine came to my dorm room, knocked on the door excitedly and said, John, come see this and I went to his room. There was a score of a basketball game on his computer and it said, Atlanta Hawks 87, Chicago Bulls 85, something like that. He said this is the score of a game that just ended a half an hour ago. And he was amazed. That was 1994. So that's when you were starting your journey. So maybe you can talk about what you've seen. You know, kind of commence with all these changes in society. How have you seen sensory evolving and changing with all the technological advances that we've experienced?
Lauren: Well, I think the main one back in 1994 was the use of computers. When I first started out in sensory science, we used to do everything on paper. When we were doing descriptive analysis which was called profiling those days. What you would measure the line the mark on the line that the panelists had done. And then we write right down all the numbers and physically measure the line with a ruler, write down the number. Then someone would check that. And then all those numbers would then be put into a calculator to work out what the ANOVA result was. Calculate ANOVA by hand within two actions.
John: Which is like good skill, you know, actually.
Lauren: Yeah, but that used to take us four weeks after a descriptive analysis study, four weeks and then we got phase, the very first time, we collected the data through phase and then we analyzed it and it took four seconds. And I still think about that now, four weeks to four seconds just because of getting the sensory science software. It's amazing.
John: It is truly amazing. And, you know, it's interesting because as things have sped up, it kind of seems to me like it's somehow harder to keep up now than it used to be even though we have all this technology. So this is something that I you know, I mean for example, at one time I mean now this wouldn't actually be an issue. But like, one of the reasons I started Aigora was actually I realized that we needed more data science in sensory. About 5 years ago, I got an e-mail from a client on Wednesday morning where she sent me a data set and said, okay, if you can get us the final report by Friday, it would be great. I wrote back and said, well, it's Wednesday. She said, yes, but it's Wednesday morning. So already five years ago. And that's with the basic things that, you know, is. But now we see this this kind of rise of automation throughout our field, where now it's almost like there's this expectation you're going to run a study in the morning and have the report done in the afternoon. So maybe you could talk a little bit about that speed change that you've seen, like the expectations from your clients or maybe what you see your clients dealing with. How do you see the speed impacting our work in sensory and consumer scientists?
Lauren: Maybe my clients are a little nicer to work with than yours because they don't normally give me data in the morning and expect the answer in the afternoon. Things are easier for me because of different data sources and different software that's available. For example, things like open source software, maybe use of R, for example, for analyzing data and also for looking at data like qualitative data from focus groups. I don't know if you've ever used RQDA to do the coding.
John: I never actually not use that. So maybe take our listeners do that. So this is it open source package RQDA.
Lauren: Yeah, RQDA. Once you've downloaded it and I use RStudio just because I'm not actually pretty grammar.
John: I think that pretty much everybody uses RStudio. The programmers are like, oh yes you can do things. There are other developing environments you might also use, but I don't know any people or programming R who aren't regularly using RStudio. So listeners get RStudio if you're not already using it. Okay.
Lauren: Yeah. It's very easy to do once you've downloaded it because RQDA actually works in the shiny app where it pops up and then you can just complete what you need to do. So you can generate terms, generate codes and then assign all the different sentences to those codes and then look at the entail.
John: So you can tag this kind of a tool for manually tagging the responses. I see. RQDA.
Lauren: Yeah. That's really changed the way I analyze.
John: Yeah. I mean it's amazing how these tools are remble. Do you know who is the author of that package? Have you been in communication with them?
Lauren: I have sent them a couple of e-mails actually when I asked him for help. I can't remember name.
John: Yeah. I mean, it's really true that once you get into this open source world, I find pretty much the authors are always responsive and helpful. There's a package I use a lot called officer. You've seen the results of this where you know, this is Dave Gohel in France, I believe. And it's a package that allows you to automate PowerPoint and word reports. And it's really amazing. I mean, you just end up with PowerPoint reports that are formatted natively, you know, in your like whatever your template is. And with the charts and the tables. And it's really when you haven't maybe, you know, the newcomers to the fields take this stuff for granted but for people like you and me who've been in the trenches, it's truly magical.
Lauren: Yes, it is. Yeah, I often say that to my children that they have it easy when they're doing their homework because they can just google the answer. We just got nothing when we were at school to look out the answer to.
John: That changes the nature of that, doesn't it? Because now it's not valuable to have information that said it's valuable to like traffic and information or something. You have to be able to take new information quickly. So maybe you can talk a little bit about that. Like, how would you compare your job now to your job say, 10 years ago. When maybe once upon a time it would have been important for you to know all sorts of things now instead you can get information. So how do you approach your work differently now within the pace of all these new technologies?
Lauren: I think the main difference over those 10 years that you're talking about is not necessarily the amount of information that you can keep in your brain. It's the amount of information that's available online. You can search for anything and find anything so you can try lots of new ways of doing things. And I think that's what's really really helpful. Recently, I've tried using the Mark Skillings technique for analyzing complicated rank data. Without that kind of information online, freely available. It's not something that you could just consider doing, you wouldn't even know about it unless you had that information available.
John: Yeah, it is really true how differently I approach other about, you know, for example is muscular procedure for, it came up and I believe penalty analysis. But I was doing some work for a client and now you can't use these obscure statistical procedures which are in fact the correct ones to use. And maybe they're not even that hard to implement. But once upon a time, it will be off the library. You know, hopefully you find you're looking for. But instead, you know, you can you can write, you can pull usually code. Now, I would advise everybody, go through the code yourself. You should have some idea what's going on. Don't just trust what you get. There was a big scan election of Python community. I don't know if you saw this, Lauren, but a hundred and fifty research papers were potentially invalidated by a bug and a python package. Yes. And one of the major python package. So I do think it is definitely good to read through the code and make sure and if you can try to rewrite it a little bit yourself. I do that in my case. And that's a good exercise for learning. But it does totally change the way that work happens. So maybe, Lauren you can talk a little bit to just kind of in that vein about the sort of work that you're doing for your clients these days and how technology's interacting with, you know, the work that you're doing.
Lauren: Well, one of the main things I work on is preference mapping which I think should be called liking mapping rather than preference thing and doing predictions. I've been working on something recently where the data that the consumers, the questions that the consumers answered. One of the questions was a yes, no question, which I can't really describe what the question was without telling you more about the study. But looking at the predictions, what I was trying to do was to predict consumer liking from instrumental data and trained panel data. And I used a machine learning technique called Weka. And I didn't know how to use that tool. So last year, I did the online course for Weka and learned how to use it and to understand how it works. Doing all the examples online and then used in a couple of projects and then actually used it in a project for real.
John: Oh, I see. So you kind of testing it. Try on doing a trial basis and then when you felt satisfied with the output, then you use it for clients.
Lauren: Yeah. You use it for clients.
John: Okay. How do you spell this, Weka?
John: This is out of New Zealand or some such or? And can you explain a little bit about this tool does?
Lauren: Well, it's open source. So it's very helpful. But what it does is it allows you to draw connections between your data. So, for example, when I was looking at the data set, I could decide the cutoff point for the yes-no based on the overall liking that the consumers gave. Well, maybe I could decide the cutoff point for the yes-no based on another measure that consumers gave maybe just about right scale or an intensity measure that consumers made. And then link all that data back and see whether you could predict whether the consumers would say yes or no. Based on the trained panel data. So that say descriptive analysis data.
John: And so this is machine learning based like predictive. It builds a predictive model for you using kind of, I guess, auto-machine learning approaches. It fits the model. Probably cross validates things for you. I would guess.
Lauren: Yes. Yeah. We're using a ten-fold replication. Very helpful.
John: Yeah. I mean, this is really like this is starting democratization. You've probably heard that. Everything is democratized these days. Like it's just as I mean, I don't know. I hear it so much, I've started to like, actually kind of tune it out. But it is true that things are getting democratized. That these advanced machine learning methods are now finding their way into all walks of life and it's very helpful. It's very powerful. You know, now we are able without necessarily going and doing in degree in, you know, whatever it might be computer science optimization, etcetera, that people who have subject matter expertise like you are able to then get access to kind of horsepower that you would not have had access to either. So how is that changed the way that you interact then with your clients or the sorts of projects? Are you able to take on new projects as a result or are you just able to reach they do a better job, if you like, an even better job than the actual job you are already doing?
Lauren: I think doing an even better job is probably the right answer for that. But it does mean that you can offer that as a service to clients. So that client particularly was interested in doing preference mapping and during the predictions in the normal way, with preference mapping. So that was all done as well, building a predictive model using the preference mapping data and then using the machine learning with the yes-no answer was a kind of added benefit and added question because I had used before. So it gives you flexibility to offer your clients to approach something in a different way.
John: So now I think a really important question here is for someone who is hearing about this and thinks, oh this sounds great. What advice do you have for how to get started? I mean, if I get that question for amount people say, oh, I really like the idea of machine learning, artificial intelligence, but I don't know where to start. So maybe you talk about your journey as you went and you found these tools, you found different resources. How is that gone for you?
Lauren: Well, I really like to see the open source courses as a nukes online. I think the Weka course, I can't remember if it was on Coursera or maybe free learning. I can't remember now which one it was. But that took me about three months to do that course. But it means that you've got all the information that you need because you've actually studied the course, download the exercises and tried it on some previous sets of data. And you can see whether it works or not. And a lot of those courses, it looks like you need to pay for but if you just say that you're going to audit the course. If you click that option and you're not interested in having a certificate, then you can do the whole thing for free.
John: Yeah. And I guess for someone like you, there's probably no reason to have a certificate even. Right? I mean, yeah, that's interesting. Okay, kind of along those lines, how did you manage to fit that into your life as a full time consultant? I mean, what do you recommend a few hours a day? Do you block time or how do you actually implement that?
Lauren: I know we do half an hour to an hour a day on something like that. As a fellow of the Institute of Food Science and Technology, I have to show my continuous professional development, my CPD. So it's very useful for that and just showing that you can use your learning in the future. One of the things that they do within the next years is looking at reflecting on the learning. So you take the learning and think about what you've used it for over the last two or three months and then consider what you might use it for in the future. So it's actually I think that's really helpful because you keep your CPD up today.
John: Oh, okay. So they want you to actually provide some input, some responses. They collect that every week.
Lauren: Yes. Well, it's still online.
John: Yeah. And so that kind of forces you do. Keep moving forward and to think about like, what you've done or you're going. It's interesting. So for the online mukes that would be one source you'd recommend for people, Coursera. I love Coursera. Actually that's where I learned a lot of my early data science was on Coursera. I find LinkedIn learning to be very good as well. Have you looked at that at all? That's good for databases.
Lauren: No, I haven't tried that.
John: Lynda.com and then they got bought by LinkedIn and now it's a LinkedIn learning. Yeah. You have to wonder, I mean so much information. What about other resources. Do you also like books and that kind of thing. Podcasts? What are some other resources you'd recommend.
Lauren: Well another resource I would recommend is joining AigoraPlus because your tutorials have been really helpful. Because they just make the whole dealing with data so much quicker. When you get a data set in sometimes people send me a data set and I look at it and I think, I don't even know what these questions were. You know, what the data is showing. It takes so long sometimes to get on top of it. And then having the tools that you've been sharing us with some tutorials just makes everything so much quicker.
John: Yeah. Such for that kind of us. Lauren, what you're talking about, I believe her that the title of our tools from that Hadley Wickham has developed a book that I've gone through this entire book. I hold it up here. Lauren and I are on video. And I was holding up my copy of R for Data Science, which is actually signed by Hadley Wickham, I dont know if you've seen that.
Lauren: Oh, cool
John: But I work through this every single line of code in this book I've typed and I have to say, it's so worth it changed my life, really. Because when you're when you have a kind of fluency in data manipulation, your life so much easier and then you can write scripts, you know, if you're getting the same sort of you never know what it is about our field, but there's so many different data formats and most of them are very strange. So it's very helpful if you have some tools for rearranging data.
Lauren: Yes. Yeah. Sometimes you get a really good agency who sends you a data set and you're kind of dreading getting this Excel file and it comes through. And each column is headed and then all the questions are on another tab and everything has been sorted and checked. And this is, oh, heaven. But other times you just get random lists of columns with strange column header and you just have no idea what they are.
John: Right. Well, you know, what's funny about that line is that, like, when you actually write I mean, you may have had this experience where I had a very strange data set recently and I wrote code to sort it all out. And then I had to tell my client. Okay. Keep it like that. Make sure it keeps coming in in this strange format. Because I have the code to dealt with it. So it's like this strange thing is now frozen in time. So I guess the frozen are goods. Yeah. So you've been talking about some of the other kind of new technologies that you've been seeing in. What are the things that you said that was really interesting is that the way that photos and videos and other kind of sources of data are starting to come into our research, maybe you talk a little bit about some of the research you've seen that's involved, not just said survey responses, but or even text responses, but pictures and videos in a kind of alternate media.
Lauren: Yeah, I think that's a really interesting area. I like the idea of the sensory ethnography. A company called Watch Me Think, for example. And looking at the way they collect the data, I think it goes down to a different layer, a different layer of detail. Because at the moment, if we are asking just a survey and you're asking people, you know, how much do you like this product or maybe propensity to purchase if you're going to ask that question. But you might ask an open ended question. And they can be very enlightening, particularly if you've got something like something I've been doing recently, sentiment analysis. And that's been really helpful. But with the photos and videos, I think that can really help, because if the consumers can come along to a focus group with videos or photos that they've made of their interactions with the products, I think that can be really good. A colleague of mine, Carol Riter, she has been working recently with a bulletin board where consumers have been uploading photos and videos. And she said to me, that's been really helpful. If you think about if you're doing a focus group, if you've got eight people there for an hour, each person got the chance to speak first a few minutes, haven't they? Right. Whereas when they're on a bulletin board, they can speak or upload videos or take photos for as long as you like. And so it just gives all of those types of approaches, I think, give you that extra detail, that extra layer of information from the consumer about why they like that product, why they don't like it.
John: Right. A kind of consumer researcher, much deeper insight into the fuller consumer experience, not just, you know, what might be kind of shallow answers on a questionnaire, but a detailed particular lives.
Lauren: Yeah, particularly if it's something like a home use test with repeated use and repeated discussion about the reasons why they like or dislike a product. You say imagine if you could follow your product to see assessed three times in a in a home situation with a family. That's going to give you a lot more information, isn't it? Than just asking people on an online survey how much they liked the product?
John: Yeah. So do you see surveys? I mean, it's interesting question is that kind of big picture, whether surveys themselves are going to either go away or undergo some very radical transformation. You know, one of the things that so my clients are interested in are alternate kind of Ritscher's surveys. So kind of Alexa based survey will be one where you have smart speaker, right? Because you just press the contacts and talked to different people who have experimented with Alexa based surveys. And there are definitely some challenges. But the idea of asking someone in their home what to think about something and allowing them to just talk and getting the answer and then processing it to digital form is very appealing. Chatbots also right? That instead of it just being, you know, what did you think? Rather noticing things about responses and then having questions that are a function of what the person's already said. I think chatbot technology is still kind of disappointing and a lot of ways. But the idea at least is a good one that we can, you know. Yeah.
Lauren: It's a very good technique, isn't it? If it could actually work and think properly about what you said before.
John: Yeah, right? I noticed that you gave us the top box on sweetness. Can you tell me more about that? You know, just some maybe even just a rule-based chatbot. But I mean, it is interesting and another thing, of course, people are realizing is that liking sports by themselves are not really that helpful. You know, you're not really getting at the kind of system wanting. So are you seeing that you know you're talking a little bit about eye tracking also, like, are you seeing your clients looking at more alternate sources of data, not just surveys or how is that evolving in your experience?
Lauren: Most of them are still on surveys, but some are moving to more towards the photography and video elements. I think it will be slow in some bigger companies. I think that some universities have the ability to kind of do that research but until these new technologies have maybe more proven. I think people are a bit hesitant to just leap in and start developing a new product just based on something like that. But I can see them becoming useful.
John: Right.Here's what all the norms are like in a you know, we're used in this product category. Good means seven out of nine or whatever. You know, there's norms in place. And so it's tough to just throw that away. But, yeah, I mean, as you know, I talked with Joanne Hort recently on this show. You know, I think she's doing really interesting work kind of setting the context. Are you seeing that in your own research as well? People are starting to play around with just alternate modes of data collection, you know, instead of just being on a survey. Like, people bring people and try to set a context. I mean, there's some good opportunities, I think, there with these new technologies to create more valid environment.
Lauren: Yeah. Particularly the holoLens, I think, because I don't think it's quite so restrictive as maybe virtual reality. If you're assessing a product to wanting to look at how something works. It's quite difficult to do like virtual reality for cleaning a worktop, for example. And if you want to eat or drink with virtual reality, it can be quite difficult. Whereas a holoLens can do that quite well. But one of my clients has done something interesting recently, which is instead of creating the content, they were working on selling alcoholic drinks. And what they did was give people try two beers. And they tried them. And then the one they liked the most, they could take the token and then they could go to the bar and have more of that product. So it was kind of like real context because it was in a bar. Right. But they could have more of whatever it was that they liked more. And the data for that was really good. It was difficult to analyze because there were lots of pairs. But it gave, I think, a much better outcome than just looking at liking data because people could choose which product. And something similar. People could choose the product to take home and use. So instead of just trying it once they could, they came along to a CLT, had three products and they could choose one to take home to use. That was a home product. So those types of approaches, I think, can give that extra detail. Rather than having to maybe go strictly to neutral technology.
John: Right. But they got indirect measures of. Yeah. Right. Working my company has done and kind of revealed preference. You know, people will do a test and then on the way out the door. Oh, we have some snacks for you. Here help yourself and then just take a note of which snacks people take on the way out the door. You know, that kind of thing. That might actually be more harmful than with the survey responses. Yeah. Kind of moving past beyond survey responses. You know, I see that in our field in general, there's been a yearning for something more informative than liking scores for a long time. And I feel like we're starting to make some progress in that area. So we've we've got maybe five minutes left. So with the time that we have remaining, I really think that your advice on how people can get started learning some of these kind of new tools is extremely valuable. Right? I mean, what is your background originally? Originally, psychologists, food scientists?
Lauren: Analytical chemist.
John: Fifty percent people of my life are chemists. It seems like I don't know why. It seems like also in data science. I would say at least half the data scientists I meet are either chemists or chemical engineers. And then for whatever reason, they go to data science. There's something about that. Yeah. So maybe you're part of that group too, because you're becoming more and more of a data scientist yourself. So, Okay, so we've talked about mukes, we talk about books. If you were just kind of lay out a path, suppose you have some young food scientist who's really interested in like, okay, I really like the idea of data science machine learning, that kind of thing. What advice would you give to that person? You know, like for her to get started with this kind of alternate palette sub, basically strengthening, you know, her quantitative abilities or her computational abilities through external learning. What would you prescribe for that person?
Lauren: Probably the first thing would be to look at some publications to determine which of these particular elements that they want to find out more about, because there's quite a lot of different things out there. Maybe look at some publications, not just papers and journals, but papers and publications in magazines and articles online, and then choose a muke, to go along with that and study that muke. But also maybe try and join some online groups where you can actually ask questions. And that's where I found a really useful contacting the people who've written these pieces of software, because people are so helpful explaining how the code works, so how you can change it to do exactly what you want to do. So it's developing this content.
John: Yeah. There's kind of a massive party going on online and, you know, the kind of knowledge sharing world. And if you're not participating in that party, you're really missing out right now. Because, like I think about the last year or how much I've learned just being on LinkedIn and following different people or I guess you're pretty active on Twitter, is that do you find Twitter to be helpful in this regard?
Lauren: Yes, very, very helpful, because you can get a lot of people are just on Twitter and you can just send them a direct message and they just reply the same day. It's really helpful.
John: Yeah. And LinkedIn, like, I read someone's book and then I get a book on Neo4j here and which is kind of open source database platform. And yeah, I just contact the authors of the book and say I like your book and I like Oh thanks a lot. And so connect with them and their connect with all these people. And it's like you can be part of this amazing thing. All you have to do is like start doing, you know, just start engaging.
Lauren: Yeah graph database, this is my next thing to learn more about because it looks really interesting. Power of Weka, is similar to a graph database. But the whole area looks very interesting for linking up, particularly for maybe developing consumer insights. I think that would really help.
John: Yeah. That is my big push for 2020. To get everybody in consumer science using graph databases like all night talk. And in every conference that I find to attend next year, I will get at least one talk on graph databases. So it's the answer that everyone wants to know, which we do through historical data. The answer is put it in a graph database that you have to have an appropriate data model, you know, do with it. But that's the answer because you need just for our listeners out there, a graph database is like a secret database, except you get rid of all the structure you don't need. So you have the relationships are just all, you take all the information granularized it down to like individual responses, individual consumers, individuals know scales, this kind of thing. Identify those relationships. And you have this massive graph. And then you can ask questions like, okay, show me the flavors that I've received Top boxes, you know, by year in this product category from this demographic group in this region. And you can just get that data and analyze it like it's honestly amazing. Like, I don't really disappointed this call. But I would say, if you don't know that graph databases that learn about them, they are what where it's at when it comes to data management. Especially for us with the diverse data. Yeah.
Lauren: I was thinking it would be quite useful for looking at clusters of consumers. Understanding more about how people's liking changes over time. If you had that information. Lots of companies have information on their databases. I thought to be quite good use of.
John: Yeah. That's great. We're shocked about this offline, but it's. Yeah. I'm on graph databases right now like it's so. Yeah. It doesn't do the analysis. Right? The database is just a way of storing information. But what a graph database does allows you to easily and intuitively get exactly the information you want. So you have some question. You think what information that I need to ask the question. You go to a database and you say, all right database, what do I already have that? What data do we have? That means the following criteria. And you lay out the things that you need. And it goes against it for you. It's like this very faithful servant that says, well, here's all this stuff. So it's great. Anyway, alright. Well, that's good, Lauren. So where can people find you? It's you know, I always enjoy talking to someone and to reach out to you connect like we've been talking about here. How would they find you?
Lauren: Well, I'm on LinkedIn and also Twitter as Lauren L. Rogers or you can email me at firstname.lastname@example.org. Don't email laurenrogers because that is the Museum of Art. Lauren Rogers and we often swap e-mails, ones for her one for me, flip them back. But it's Lauren L. Rogers. Just search for that online and you will find my website as well.
John: We'll put all the links in the podcast notes.
Lauren: Thank you.
John: Okay. Well, it's been great, Lauren. Do you have any kind of parting advice for listeners? What do you like the next, say, two years of hearing and give someone advice for what they should be doing? What would your advice be?
Lauren: Well, the thing I'm working on a lot at the moment is data visualization. So try and to show the value of sensory and consumer science data through better data visualization. So I'm giving a talk in a couple of weeks with the IFST about that. I gave same talk at Pangborn in the sensometrics parts. Yeah. But it's a bit jazzed up for Christmas time. So some mince pies and various things like that to be consumed. But I think data visualization, if we can share better what we do, the outputs, for example, from preference mapping, things maybe sometimes too complicated for people to understand. If we can share them in a more accessible way, maybe sensory science will be even more popular than it is now.
John: Right. And have a larger impact and a bigger impact. Yeah, that's great. Okay, Lauren. Well, thank you so much for being on the show. It's been great. And I look forward to seeing you in person conferences next year.
Lauren: Yeah. That would be great. Thank you, John.
John: Thanks a lot. Okay, that's it. Hope you enjoyed this conversation if you did. Please help us grow our audience by telling a friend about AigoraCast and leaving us a positive review on iTunes. Thanks.
That's it for now. If you'd like to receive email updates from Aigora, including weekly video recaps of our blog activity, click on the button below to join our email list. Thanks for stopping by!