Live code along: How a data science portfolio can help you land your dream job

Hello and welcome everybody to today's Live Code Long. We will start now. So in today's session, we will show you how a data science portfolio can help you land your dream job and how workspace can help you build a portfolio. As a reminder, the session will be recorded. We will short the recording afterwards via Slack email and via our social media platforms. So, first of all, thank you for attending the fifth Workspace Live Code along. A huge thank you as well to our speakers, Jen and Richie, who together with all of you and datacamp made this event possible. So next slide please, a little bit about datacamp itself. So our mission statement is to democratize data science education and make data literacy accessible to millions of people and businesses around the world. This is done through our learning platforms and now also through our Data Camp workspace certifications and events such as this Live Code along our speakers today are Shen and Richie. Shen is the head of Career services here at DataCamp, and Richie also works at DataCamp as the learning Solutions Architect. Today's Live Code along is divided into several parts. First, Zhen will explain the importance of having a data science portfolio and why it is really important during your job hunt. Next, Rich will take the floor and he will show you different ways of finding an effective data set. These first two parts will be concluded with a short Q and A where you can direct your questions to the speakers themselves. In the second part of this live Code along, Ritchie will go over an example data science publication that he made using datacamp Workspace. In this part of the session, he will analyze the data set and then Rich will also explain how you can use datacamp Workspace to create your data science portfolio. Also, this part will be concluded with a short Q and A, so if you have any questions, feel free to ask them via the question button on the control panel of GoToWebinar. Richie and Jen will then try their best to answer them during the corresponding Q and A sessions. So thank you for your attention. And now into the Live Code along itself. Over to you, Jen.

Thanks so much, Jenna. Welcome, everybody. As Senna said, I'm the head of Career Services here at datacamp. What that means is I essentially work directly with any learner who's gone through one of our certifications. We currently have certifications both for data scientists and data analyst professionals. And once they complete the certification, they get to work with me and we discuss their career goals and help them prepare if they're currently looking for work and and talk a lot about personal brands. And so today, obviously, I'm Going to focus in on the portfolio piece of your personal brand. Because portfolios are really the one way employers have to judge your technical ability, where maybe everything else is more of a description of either learning or work experience that you have. Let's go to the next slide. So here's sort of this interesting conundrum. When it comes to portfolios. You have to have a portfolio that goes without saying. Anyone applying for any sort of data role really should be prepared to demonstrate some of their work. The thing you have to keep in mind is some people may never actually look at your work. They just sort of check off the box that you have it. So it's kind of a lose lose scenario because you invest so much time usually in building your portfolio. But some hiring managers don't ever really look deeper than the fact that you have it with some titles and maybe a one sentence description of the project. So what I want to encourage you to do is think about how do you get them into the. Into looking at your full portfolio? How do you make it interesting and engaging enough that they want to see more than a title on your project? So, next slide. So why do you need to have your portfolio? For one, as I said, it really is the best means of showcasing your technical ability. Any data role really falls in the gap between a business professional and a technical role and you get interviewed on both qualities. So you really want to make sure that you have something that can showcase your work to the best of your ability. You want to get noticed by hiring managers. Creating good, unique, interesting portfolio projects will allow them to get insight as to who you are and how you work. And then if you're looking for some specific industry for your next role, it's great to make sure you have portfolio projects that demonstrate your domain knowledge. Show them that you know the type of analyses that you'd have to do with that organization or company and show them you're ahead of the curve there. And then lastly, it's a great way to engage with the data community. Share your work out there, publish it. Richie will go over this with you guys, share your work and start becoming part of the conversation. And just instead of watching all of the blogs and and Twitter feeds that show someone else showcasing their work, let's put you in the limelight because the more you build your data community, the easier it will be when you're looking for your next role. Next slide. So here's just a couple of tips to get you started. Richie will go through a lot more on the technical side, but I just Wanted to emphasize these because they're really essential when you're thinking about a portfolio for a job search. Use a unique data set. Take a unique approach to a data set that maybe has been used before. I can put it out there right now. No one wants to see another Covid analysis as part of a work portfolio. It's great for practicing working with data, but maybe it's not that interesting for a hiring manager. Find something that's unique that will help you stand out from the crowd really does make a big difference. Focus on the quality of your work, not the quantity. I'm not going to ever give you a specific number. You have to have five projects. It doesn't matter how many projects you have if they're not very good in terms of quality. So let's make sure you're showcasing high quality work instead of making sure you're ticking the box for how many projects you have. And then lastly, just to reiterate, publish your work, put it out there, be part of the conversation, get feedback from other data professionals. It really will help you grow your network, which is one of the most difficult things to do when you're looking to get a new job advancing your career. With that, I'm going to hand it over to Richie and talk about how to find some of those unique data sets.

Sure. Thanks, Jen. All right, so the first way to find an interesting data set is just to use one of the data sets that we have built into datacamp Workspace. So I'm going to be showing you this in more detail later on, but at least to get started, maybe with your first publication, I would say don't spend hours like trying to find a data set, just use one of the ones that we have to begin with. And that's going to get you practicing just analyzing data and sharing it without having to worry about have I found the perfect data set or not. Once you do go beyond like your first couple of publications, there are a ton of resources around just for finding data sets. So in particular there are quite a few places that just curate interesting data sets. There are lots of data sharing platforms and beyond that most governments will have some kind of legislation around. If you're a government entity, then you need to make a lot of your data publicly available. And so particularly governments, local authorities and non governmental organizations, there's going to be a ton of data sets around that you can go and analyze so you can get the data for free, download it and just do your own analysis. There are loads and loads of different data sources actually One of my favorite things on this page is the Data is plural newsletter. So you subscribe to this email list and every week you get an email which contains half a dozen interesting data sets. So that's like the data comes to you, you don't even need to go out looking for it. So yeah, plenty of ideas for finding data sets. This list is really not in any way comprehensive. If there's a particular type of data you're interested in, Google search is good as anything. Just type some interesting keywords and then data sets and see what comes up. And beyond that, if you can't find an interesting data set or there's like something you're really interested in, you quite find that with like academic papers and things. They might not provide the data set within the paper, but if you email the authors, they're possibly happy to send you the data set to work on. So don't rule out the option of just like asking people for data. I'm sure like a lot of people, they want to make their data sets more widely known and so they're often happy to share data with you, sort of pending legal restrictions. So go hunting, see what you find. If you can't find something, then just ask people for data works really well. Okay, so that's the end of the slides. I think we're ready for the first Q and A session.

I see a first question here about participating in Kaggle competitions and whether or not that's a good way to start building your portfolio. I think it's a great way to get practice and start engaging in the data science community. It may not be the perfect project to demonstrate all your abilities only because the feedback I've received from hiring managers is that there's so many solutions to Kaggle competitions out there in the ether that they have a hard time knowing for sure that it was your unique code. And so it may not be showcasing your specific technical ability. So great for practicing, obviously wonderful for modeling. Definitely take part in competitions, whether on Kaggle or Data Camp or any other platform that offers data science competitions. Just again, think about it from a uniqueness standpoint when you're talking about a portfolio for job searching.

One thing I'd like to add to that is Kaggle. The community is very much focused on machine learning. So if you want to get a job that has a lot of machine learning involvement, like a sort of machine learning learning scientist or machine learning engineer type role, then yeah, Kaggle's pretty good for that. But for other data roles, particularly like data analyst roles, data Scientist roles. The Kaggle community sort of isn't really as focused on that sort of thing. So yeah, I recommend Data Camp all the way in that case. So the next question I see on that. So do you have any advice on the appropriate size of a data set for. For a good project? And that's a great question. So in general, no hiring manager is going to care about the size of the data set you use. I'm going to get into this in more detail in a moment, but in general, when you're hiring, you're not going to spend hours and hours looking at someone's portfolio publication. It's going to be a very quick thing. Like if I'm hiring, I'm going to spend five minutes looking at someone's publication maximum and probably I'm going to want to try and determine from this publication does this person have the requisite skills or not in under a minute. So don't worry too much about the size of the data set. In general, you're going to want to keep things simple if you are writing a book length analysis or something. I'm not going to read it all when I'm trying to hire you. So in general, yes, smaller and simpler is fine and arguably better than some sort of large complicated data set. Okay, the next question.

Is.

Can you do a group data science project? I'm not totally sure on the, on the exact sort of details of this question. So whether you're referring to is it possible to do a data science project as a group and then use that in your portfolio, I would say yes, you can showcase a group data science project, but for hiring purposes I'd want to know which bits of the work are yours rather than someone else in your group. So you might need to amend it to only showcase the portions of the work that you've done. Just highlight I did this bit. Or you might want to just think about having a project that's solely your own work. I realized if it's a large project, so particularly if you think about how academic papers work, if you publishing something in a journal, very often you have multiple authors. And so it's fine to showcase that you're an author on this paper, but the hiring manager can be clear that it's not all your own work. So you're going to want some way of showing, hey, this is the bit I did. All right, so next question is, do you need a premium membership to access the data sets on Data Camp? So the actually all the data sets are available like workspaces available to Everyone, the data sets come with it. I'm going to talk you through the data sets in a bit more detail, but in general actually let me just show this on screen. So if you have access to workspace you should have access to the templates page. So I guess if you go to log into Datacamp, go to app.datacamp.com, click on workspace and templates. You should see this. If you can't see this, then you're either not logged into datacamp or not registered yet. But you should. Everyone should have access to this. So next question. Does it add a lot of value if the project is deployed in Flask Docker or STH and not only in Jupyter? So that really depends on the kind of job you're going for. If it is maybe more of like an engineering type role and the job application or the job description actually asks for flask skills or docker skills or whatever, then yeah, you're probably going to want to showcase those skills for most data analyst roles, most data scientist roles. If the, if the job is all about analysis, then Jupyter is absolutely fine for most organizations. Again, it goes back to have you read the job description and are you demonstrating the skills that are listed within that job description?

Richie, I can take the next one here. We got a question about what type of data is good data to use for a portfolio. And this person says they're coming from academia so they have more survey based data. They don't necessarily have industry based data analysis to showcase. I would say probably find a means of translating the type of analysis you're doing through your research and translate it into something that's using data from the industry you are target targeting. If you have a specific industry or the type of role that you want to do, it will help your chances for sure. Because there is a perception that sometimes folks coming from academia finishing their PhDs maybe don't have any practical experience and some companies aren't necessarily wanting to take on the role of teaching people how to work in industry based job. So the more you can demonstrate that they don't need to help you translate your skills into industry, the better off you would be.

Okay, so there are three more questions people have asked, but they're all things that I want to cover in the next section. So I think maybe if we move on and I'll talk through. Oh actually hang on, there are a lot more questions so we can skip the next three questions. I'll talk about this in a moment. The only one actually. Yeah, Maybe we'll come back to all the rest of the questions later in the, in the next Q and A. If I've not got to them, I think most of the stuff I'm going to cover within the next section. All right, so let's skip over to a publication I made earlier. And so I'm going to cover some of the sort of more practical details of creating a publication for your portfolio. Now if I'm hiring a data analyst, there are maybe four things that I'm going to be looking for. I want to know can this person actually do data analysis? And so by data analysis I mean can you import data, can you manipulate it, can you visualize it, can you perform computations on that data? The second thing is can you write code? In this case, like for a data analyst role is going to be can you write SQL code, can you write Python code or can you write R code? It's usually going to be one of those three. Now I'm aware some organizations, a data analyst would be using business intelligence tools like Tableau or Power bi. If that's the case, then you want to be able to showcase your abilities in those technologies. But for today I'm going to focus on Python and SQL. And the third. Oh, sorry. The reason I want you to be able to write good code is basically that if you are well, so any code you going to write over the first few months, particularly when you join a new job, is going to be me as your manager reviewing that or maybe someone else in the team reviewing what you do. And if you can't write readable code, then it's just going to cause frustration because it can be a lot of back and forth about what do you actually mean with this? The third skill I'd be looking for is can you write a coherent sentence? And the reason for that is if you are a data analyst, you're going to be writing reports regularly and other people are going to have to read them. So if you can't write sentences properly, then again it's going to cause frustration and you're probably not suitable for the role. The fourth thing is can you communicate results to a non technical audience? So when you're doing data analysis it is very easy to get stuck in doing in just writing technical language that is going to be readable by other data analysts or other data professionals. But often, as Jen mentioned, it's really important to demonstrate the fact that you can write for a non technical audience and understand what the problem domain is. Now I focus very much on data Analysts here. These principles are applicable to most other data roles as well. So if you're a data scientist, statistician, business analyst, machine learning scientist, whatever you need to think about, like what are the skills you need to demonstrate? And it's usually some variation on those four. If you're a machine learning scientist, it's rather than showing that you've got data analysis skills, you're probably going to want to show off that you understand machine learning. You can run a model, you can make predictions, all that sort of thing. So one thing to note about.

The.

Publications you need in your portfolio, if it's just for a general purpose, networking, showing off your skills to the wider world, choose a data set that is interesting and just, well, is interesting to you, interesting to a broader audience and that's going to help you do networking and sort of self publicization like through social media or whatever, it's going to build up your personal brand. The publication I've created here is a more targeted thing. So if you are applying for a specific job, then you probably want a publication that demonstrates specific skills that are required for that job. So this is an example of like a really targeted publication that's going to help you get a specific job. So as I mentioned, if I'm a hiring manager, I might get like 50 people who from their resume and their covering letter seem like they are technically qualified for the role. But the reason I'm hiring is because my team is understaffed. So I'm busy. The rest of my team is busy. I do not have time to do a 30 minute technical interview with 50 different people. So I'm going to be very aggressively trying to find reasons for not bothering to interview people. And the job of this publication is going to be to demonstrate that you have the requisite skills for the job as quickly so the hiring manager can do them as quickly as possible. So I'm going to be looking at this from a hiring manager point of view. I'm going to be looking at this for maximum five minutes, probably one minute tops. So in this publication you have one minute to demonstrate to me that you have the skills needed in the job. So from your point of view this is actually a benefit because you can make these analyses really quite short. So if we just scroll down through this, this is like it's a few pages, but it's mostly just output. So actually you can read this in a minute or so and you can get a good feel for the sets of like skills much faster than that. So let's get into Some of the details. So first of all, this background section, really, really important. So in this case it consists of four sentences. So the first sentence is just what's the data set used? What's this analysis about? So very quickly and go, okay, we're analyzing sales data. I chose a sales data example because sales data is broadly applicable to many organizations. Almost every company is going to be selling something. And so sales data works really well in a lot of cases. Avocados, I'm sure most people don't care about either way. So the next sentence describes what the analysis covers. And actually this is like a really sneaky way of just stating upfront which skills you are claiming to have. So in this case I'm saying I can retrieve data from a database, I can do data analysis tasks like data manipulation, computation, visualization, and I can answer business questions. So I'm stating that I have these skills. And the rest of this, this publication is just demonstrating that that's actually true. So these next two sentences, because most job descriptions for technical positions are going to request that you have skills in particular technologies. So here, right up front, I'm saying I can do Python and SQL and then getting into more depth, it's saying I have skills in these four Python packages. So in this case it's NumPy and Pandas for data manipulation, SQLite 3 for out 16 databases, and plotline for data visualization. And of those, I think Plotline is maybe the most controversial one. So I'm aware that matplotlib and Seaborn are more popular or at least more widely used data visualization Python packages. I tend to use plot 9 because it's a clone of the R package ggplot2. And because I spend a lot of time switching between R and Python, it works really well for me just being able to write the same code in either language. So if you are applying for a job at a company and the job description says we use matplotlib, then yeah, I'm very sorry for you. Maybe don't apply there. I'm kidding. So matplotlib, if they want matplotlib, then you're gonna have to demonstrate that you have matplotlib skills. Draw the plots in matplotlib, easy as that. A lot of this, like first section is thinking about like what are the job, what are the skills actually being asked for in the job description? Make sure that you demonstrate those in, in the publication. That's great. I'm glad that people are commenting on this. Yep. So this, this publication does have Public comment access. So if you if you have a question about any of the sections of this publication or you want, or you have any comments on it, then feel free to use the Comment feature. So just to the right of the cell, click on Add Comment and you can write your message. And we'll go through those at the end in the final Q and A. All right, so first Cell is just importing the requisite Python packages. This is another way of just like repeating what you mentioned here. I think it's really helpful to do all the import statements right at the top. This is going to give whoever is reading this context on like what technology you are using right up front. So like in this first sort of 15 seconds of reading through, you've said what's the data being used? What skills do you have? What technology are you using? And it's all really clear. Moving on to the next section. In this case I stored the data in a SQLite database. Now the Avocado sales data is really, really simple data set and it's just one table. So I could easily have used a. CSV file for this. However, because a lot of data analyst roles will ask for like SQL skills plus Python or R, you don't want to have to write two different publications to demonstrate each technology separately. Because hiring manager again, they're going to spend one minute trying to figure out do you have the skills or not. They're not going to have time to read two different publications. So just by storing your data in a database and then retrieving it, you get to show if you have SQL skills and then you get to analyze it in Python RR and you get to show off that you have skills in two technologies at once. The reason I used SQLite is that SQLite is a database type designed for small data sets. For example, your web browser is going to use a SQLite database to store the cookies in. For example, SQLite databases are a single file, so really portable. It takes two lines of Python code to turn a Pandas data frame into a SQLite database. Makes it really easy to upload into workspace. The setup of a SQLite database is really simple. It makes it ideal for this sort of situation if you want to show off your database skills in more detail. Like I could have created an AWS account, set up a postgres database on there, and then imported the data into the workspace from postgres. That's a bit more effort if you're applying for a job where you want to show off you your database skills. A lot and show off your cloud skills, then absolutely do that for minimal effort. I would say SQLite is the way to go. And so in this case the first thing we want to do is explore the database. I think exploring a database is really good, sorry, a really good skill to show off just because it's something everybody has to do at some point. So in this case it's two lines of code. We create a connection to the the SQLite database in this case, because SQLite database is just a file, you pass the file name and then we're using the pandas read SQL method, sorry function to explore the database and SQLite databases. They'll store the list of tables in another table called SQLite Master. In this case you can see that there is only one table in the database called Avocado Sales. The next thing we have to do is look at what are the columns within that table here. I really like this line of code because it's a big show off thing. SQLite databases have these extensions called pragmas which are just shortcuts for doing common database administration tasks. In this case you can type pragma table info and the name of the table and it will tell you, okay, there are seven different columns. Those are the names the data types in each column are null values allowed? What's the default value? Which one's the primary key? I think because pragmas aren't that widely known, they're considered to be maybe an advanced database skill. But actually it's just one really simple line of code and it's something you can easily remember and reuse from one project to another. So it has the benefit of like letting you show off. But also it's something that's actually really easy. So here already I've demonstrated, I have some at least intermediate level SQL skills and it's just three lines of code, really straightforward. So the next thing is I need to pull out the data from the database. Again, we're using the Pandas Read SQL method. When I write longer chunks of SQL code rather than passing it directly into Read SQL like I did before. Like here it's really helpful if you can write SQL code that's sort of correctly formatted, you're going to get less bugs that way. It's going to be easier to read and it demonstrates that you can write readable SQL, which again, one of the things I mentioned was like you need to be able to write readable code, otherwise no one's going to hire you. Because who wants to write? Who wants to review Someone else's garbage code, all that. In this case I'm using the Python triple quotes to have a multi line string. I can write my formatted SQL and I store it in a variable called query. This pattern of storing SQL code in a query variable and then calling read SQL, just passing that variable in. This is a really common code pattern you're probably going to want to use quite regularly. We've just pulled the data out of the database into a Pandas data frame. Now the next thing is we want to get straight into answering business questions. And the way I like to structure these sort of analysis, sort of for your portfolio, like when I'm like for hiring purposes, you want to make each section header the name of like the business question that you are answering. That way it's really prominent that your analysis isn't just some sort of noodling about with data, it's really answering actual sort of business questions. So the next thing is it's I've included this as a slightly annoying thing. That plotline needs a little bit of data manipulation in order to make the plots wanted to draw work. Normally I wouldn't necessarily bother including these sort of like minor details within a publication, but for the purposes of like a portfolio where you're trying to show off that you do have data manipulation skills, I think it's actually a bonus to include this sort of thing. So in this case I need to create a new series within a data frame from two other series, some demonstrating, oh yep, I have a data manipulation skill. And then also I just include this line just to show the different values that are included there. So this is just demonstrating some of my PANDAS vocabulary. So in this case I can use the unique method and the to list method. So again it's not crucial to the analysis, but it's showing off that I have some Pandas data manipulation skills. And then we get on to drawing plots. Now I wanted to draw a bar plot in this upfront because bar plots are one of the most common plot types used within business analysis. So when we're analyzing a sales data set, any kind of like sales data analyst role, I'm going to go for going to be expected to draw bar plots. So just including a bar plot right up front, right up front within the analysis is really helpful. The other thing to note is that I've included a description of what the insights are from this plot. So again it's not just like can I draw a plot? It's can I interpret that plot and provide human readable explanations of what's going on. The other trick I've used here is that because hiring manager is going to be skimming through this again, it's like they're going to want to spend a minute or less reading this. I've highlighted keywords in bold. No one wants to spend the time reading two sentences to see if they're interesting or not. But someone will skim read bold text, and that way they're going to get a sense of like, oh, yeah, this is important. We've got some insights from this plot and really, really easy and quick to understand what's going on. This code here, this is just plotline code for drawing a plot. One thing that's worth noticing is that I've made heavy use of the Python backslash for having continuing code from one line to another. So Workspace Publications, we've taken a lot of inspiration actually from a blogging site called medium.com and there they've done a lot of research to suggest that when you have, like, narrow publications, people tend to like, read more and read faster. So you'll notice that workspace publications are quite narrow by design and this is designed to stop you having to look from side to side so you just like, you read faster that way. And so one consequence this has is that because Workspace Publications are fairly narrow, your code needs to be so narrow and long rather than super wide and having like 200 line lines of code. So I think this makes it more readable if you're not familiar with the Python backslash operator, to have continued lines of code. Well worth getting a feel for it using that. So the next business question I wanted to answer was, well, it's less about the business question in this case. It was more that I wanted to draw a line plot. So line plots maybe the second most popular type of data visualization used in sort of business analysis. So I wanted to show off that I could draw a line plot. In this case, I made it fairly fancy and we've got a couple of different facets in there using color and legend. Again, I have listed all the insights from, from this plot. I've highlighted words in bold again and again. You can see like, this section is named after the. The business question I'm trying to answer. So one thing that came up when it, when I drew this, I was like, well, actually there's some seasonality in the plot, but you can't really see exactly what's going on. So although I was like, well, you know, I've drawn two plots. That's probably Plenty for someone who's really skim reading this, because this question came up, it looked quite interesting. I was like, okay, maybe we need one more plot just to show off that I can create a narrative of like asking questions, answering them, and then that generates new questions and then I can go on and answer more questions. So the next section I was looking at seasonality. In this case, rather than looking at the date on the x axis, we're showing date of year, sorry day of year. So that's like number From January the first is one up to December 31st is going to be 365 or 366. So in this case showing off a bit more sort of data manipulation skills. And then I've got this other plot, so I get to show off a few more things. Like in this case I'm drawing smoothed trend lines for each of the the cells, each of the panels within the plot. And you can see here that in fact Organo, particularly with the organic avocados, we sell more of them between February and June and then for the conventional ones you get to see, well actually in winter as well. And again we've got all the, all the kind of key insights from this plot listed and they are keywords highlighted in bold. The final thing I realized after I've done all this, I've shown off my data visualization skills, but really I need to show off some data manipulation skills as well. So I threw in a pivot table at the end just to show that I can do computing on data as well. And all in all, again, couple of minutes to read it looks good. And I believe I've shown off all of these skills that I listed. All right, so let's switch over to workspace itself and I can show you how this was sort of made. So if I go to my workspaces and I'll click into the one I've made already. Now workspaces consist of jupyter lab sessions. So on the left hand side you can see the got a file browser and there's a few sort of menu for commands. But the bulk of the screen is a jupyter notebook. So these have the file extension ipynd. So they used to be just called ipython notebooks, so that's why you get that funny file extension. But here you can see the contents of the notebook look pretty much the same as the publication. So we have text cells and we have code cells. If I click into one of these, you can see how they look. So in this case the text Cells, they're actually markdown text. So it's basically, it's normal text, but you have things like you can write things in two stars to make them bold. You can include hyperlinks using this syntax and so on. And if I do SHIFT Enter, it's going to run that cell and show you how it looks for the end user. I can also run the code cell by clicking Shift Enter. In this case, it's going to run the cell. Actually, that one's not so exciting. Maybe if I run this one, you can see what it looks like. So I run the cell and it's going to show the output in line.

So.

The way you're going to do this, you're going to just write all your analysis and then once you're done, you click share. In this case, I've published it once already, so I'm going to click update. But to begin with, it's just going to say publish. This is going to run every single cell within your notebook, and once it's done, you can open the link and this is going to show you the public version, which you have seen already. So, one other thing I wanted to show on this particular workbook, there's actually a hidden cell here. So if I click into this, one of the first things I want to do when I'm creating a publication for my portfolio, particularly when it's something specific to a job, is I want to write a note to myself on, like, what skills am I demonstrating? And this is something you can pretty much copy and paste from a job description, but it's better to be clear on, like, what does the analysis need to cover before you. Just like, you write something and then realize it doesn't match up to the job description at all. So just keeping it clear in your mind, like the things that you need to demonstrate is really, really important. So. So in this case, it's like, yep, I want to retrieve data from a database, calculate some summary statistics, do bar plots, line plots, and I want to make sure it's good idiomatic Python code. I don't want to have good sentences. And so one thing about this, because it is a note to self, I don't want it to appear in the publication. You can see you've got this angle bracket, exclamation mark, two dashes at the top and then two dashes close angle bracket. This is the syntax for an HTML comment. So you can wrap whatever things you don't want to appear in the publication within these HTML comments, and then you can see they don't show up here at all. It's completely invisible. So this is a really great tip if you just want to write some commentary to yourself, make sure that you're clear on what you're doing and not. And not have to worry about, like, the reader seeing it. All right, so that's everything I wanted to show you about the avocado sales analysis. I do want to spend some time going through the workspace templates. So as I mentioned before, this is a really good way to get you started, particularly like with your first couple of analyses just so you can show off something really quickly. And so if I click on data sets, each of these data sets is designed to be sort of broadly interesting. And so even if you don't find like one that's like perfect, I think like, a lot of these, like, form a great sort of starting point for like, showing off your first analysis. So for example, if we click on E commerce data, you get a preview of what the data set looks like. And so, yeah, it's going to show you what's in the data and give you some examples of interesting questions that you might want to answer with the data set. So I'm going to click Use template and show you what happens. So this is going to create a new workspace for me and they should have a notebook with things in there already. So basically edit this as you see fit, start coding, start trying to answer questions that you find interesting. Now, because this has no analysis in at all to begin with, I would say this is for slightly more advanced people like you. You're going to have to have some understanding of like, what are going to be interesting questions to answer to begin with. So the other thing I want to show you are the recipes. So recipes will include ideas of particular skills. So for example, if you want to be able to like, demonstrate that you can find and visualize clusters with K means, let's create this template. So, So this is going to give.

Me.

A workspace which tells me, like, all the things I need to do in order to do a K means analysis. So if you want to know how to like, like a good way of sort of demonstrating particular skill, you can use these, these recipes as inspiration. The third thing are the playbooks, and these tend to be a bit longer. So these have, basically they have a data set, but they also include the analysis within them. So rather than just giving you a data set and saying, okay, just go work on this, it's going to, it's going to have like a more complete analysis. That you can read through. So I would say all of these things form great starting points for your first publication. Excuse me. One thing to note is that for your own portfolio, I would say do not just use one of these templates and claim this is your own work. That's really not going to fly with any hiring manager. You can get found out pretty quickly. I'm not sure what's going on there, but you do need to make sure that you just use these as inspiration rather than just the final publication. So change the code as you see fit, write your own code, do your own analysis, and that's really going to help you show off your skills. All right, I think that's everything I wanted to say on Workspace for now. So maybe we'll go back to the Q and A. All right.

So.

The one of the questions I missed out from the previous Q and A was it had been asked by a few people was just what skills are necessary to demonstrate. As I mentioned, really if you're trying to create a publication for a specific job, then those skills, they're going to be listed in the job description. So make sure that your publication checks all those boxes on the, on the job description and shows off those those particular skills. The next question is, is it better to work on projects from different domains or only focus on one? So this very much depends on like, what your sort of purposes.

So.

If you have a really strong idea of like, you know what, I want to be a marketing analyst, I only care about marketing, then yeah, you're just going to want to create marketing publications. So yeah, if you have like a specific goal in mind, be focused on that specific goal. If you are just generally hunting for like some kind of data role, then you want to be a bit more varied. It's going to give you opportunities to show off the breadth of your skills. So it's like, well, I don't really know what I want to do with my life. Maybe I'll just like create some publications on a variety of subjects, just whatever interests me. And then that's showing I can analyze a wide variety of data sets. So again, like the kind of publications you create, it just depends on your goals and how tightly focused they are.

I'm going to take the next question. There's a number of different folks who've asked similar versions of this question. But how to demonstrate your projects on your resume and should they be customized for every job that you apply to? Also there was a question. If you have a lot of projects that are varied, do you Share all of them or just some of them. And what I would say to all of these is that you should customize your resume for each job you're applying to and really choose maybe up to the three most relevant projects that you've done so that job. That being said, your profile page here where you can demonstrate all your public shared projects. You can include everything there. If they're intrigued by what they saw for the first three, then they can take a glance at your other projects if something stands. Also with how to post your project on your resume, you should include the title of the project and then one to two sentence synopsis that highlights what the problem was, what the business question was that you were trying to solve, and what technologies were incorporated in that project. And then I would typically say use a hyperlink in the job title. You don't have to write out the URL, just make the title clickable. PDFs go directly from the PDF to a website so you don't have to spell out what the URL is to the particular project. You can also include your profile page for workspace or GitHub or your personal website up on top where all of your personal information is housed if you want to just put everything out there. I think that covered a number of different questions we got.

Okay, so there was a question about how do you, if the industry doesn't use open source software like R, how do you best showcase your skills or keep skills up when not using those regularly? Okay, so if, if the company is using non open source software then okay, at the moment datacamp is not going to support that. But there are ways you can publish things. So I guess the most common situation is going to be they're using some kind of business intelligence software. So both Tableau and Power BI have ways of publishing any dashboards you create. So I'd say use either Tableau's or Microsoft's tools for like publishing your dashboards. That way if the company is using sas. I am not sure what the best way of publishing SAS analyses is. I presume it exists. I've been trying to forget about SAS for a long time. So yeah, I'm not sure what the best technique is for that, but pretty much every technology you're going to use there will be some way of, of publishing things. Data campers focused on Python and R because almost every company is going to be using one or the other at this point for analysis and data science tasks. There are a few exceptions and there are going to be some roles where, yeah, technical skills Aren't like using R Python. Aren't can be requested. But yeah, for the majority of analysis and data science roles, you can be wanting one or the other. Okay, so next question is how should we strike a balance between brevity and explaining what we did in project, especially when it comes to modeling the data. Yeah, so that's always a challenge is how do you keep things concise and yet show off all your skills. And if you're doing some kind of complex analysis, complicated analysis, then it's going to be, yeah, maybe you do need to take more time. One thing that's possibly worth doing is just having different versions of things. Again, it's like Jen said, like tailoring things to a specific job. So if you just want to show off your machine learning skills, then maybe you want something that covers just the machine learning part of things and have a separate publication on the data analysis part or the data cleaning part and then just link to that one shorter bit which you think is the crux of the job description. I think with job descriptions quite often, like the hiring managers, you see a list like 10 different skills, but actually if you read the job description carefully, there's like only like two or three of them that they really care about and those are like the ones you're going to want to focus on in the publication that you're demonstrating for a particular job. If it's a more general publication where it's just like, yep, I'm interested in this, then I would say in general shorter things get read more. So again, like sometimes you can try and like gloss over some of the more complicated details. If it is a really technical scientific report and you want people to be able to reproduce what you've done, maybe you do need to go into more depth. So it's the standard thing with, with any publication is just think about your target audience, what do they want to get out of it? And just try and judge what you write based on that. I mean, oftentimes if something is really long and just verbose, you can be quite aggressive in your editing and particularly of like how you edit your code. Just try and work out. Is there a quicker way of demonstrating the thing I want to demonstrate? All right, we've got a ton more questions and struggle to help. I realize we're coming close to time now, so what's another good question.

Richie? While you're looking at the technical questions, I'll answer something very quick. There is a few questions about career services and accessing them and who we work with. I just Want to clarify? We only work with folks who've gone through certification, completed their certification, and it doesn't matter where in the world you're located. Career Services works with all data camp learners who've gotten certified. So just want to put that out there.

Okay, so another technical question. So when doing these projects, should there be an overall question we should be trying to answer or does it not really matter? For example, in this workspace there's lots of little questions but no conclusion. I would say if it's something for like going for a specific job where you just want to like demonstrate you have these specific skills, I would say an overall conclusion doesn't matter too much because again, the person is going to be skim reading it. They don't actually care that much about the analysis. They just care like, do you have these skills? Like can you check the box that you can do data analysis? Can you write code? Can you write coherent sentences? Can you explain things to a non technical audience? So the actual conclusion is like, well, they might discuss it with you briefly in a job interview. I would say it doesn't matter that much for like a real world report on something. Like in an actual job you probably do need to be able to write conclusions, but for this I would say don't sweat it too much. All right, so another technical question. Do hiring people understand that if I know Pytorch, I'll learn Keras quickly, or if I know Seaborn I could quickly adapt to matplotlib? Or they don't care and I have to be 100% accurate like the job description. So if they've written down specific technologies in the job description, you might as well demonstrate that you can use those technologies. If the job description just says you need to be good at deep learning, then it doesn't matter whether you use Pytorch or Keras or tensorflow or whatever. Yeah. So again, go back to the job description. Treat that as like the canonical list of skills that you need to demonstrate. There's just a question about is it possible to link the data workspace? Sorry, is it possible to link data count workspaces to GitHub? So the answer is yes. So if I open up. Oh, actually, sorry, from here, from the main page, if you have a GitHub repository, you can create a workspace from that GitHub repository. So yes, they can be linked. There's one more question on data sets. So in terms of choosing a topical data set, is it better to choose something innocuous like Avocados or Something memorable but potentially more controversial. Okay, so in general, most organizations don't like controversial people because they cause problems. And then there's a whole mess with HR and it's not going to be helpful. So you probably want to stay away from doing anything too controversial. I mean, this isn't true of, like, every organization. Some organizations do enjoy having, like, more political members of staff. So think about the culture of the organization that you're applying to work for. I would say in general, you don't want things that are not suitable for work. Yeah, don't do like, an analysis of, like, the top hentai sites or whatever. You're just gonna, like, get rejected outright. Do keep it really, like, clean, safe work. I would say in general, most of the stuff that's going to be like, applicable to, like, the job that you're going for, it's going to be relatively uncontroversial. Like, I don't know what exactly you were thinking about was controversial, but, like, if it's like some radical ideas for how to improve a sales pipeline, then maybe that's going to be, like, acceptable. In general. Yeah, I would stick to something that's not going to annoy the hiring manager and get you rejected outright.

Can I just follow up for one, One other second? It is important you name the source of your data. So if you're just getting it from a template, just make sure it's clear it's coming from datacamp if you happen to remove it from your workspace. Naming your source for your data is important just to have the accreditation. So don't forget to do that piece as well.

Yeah, so I forgot to mention this when I was talking about the start of this. So this just talk about it gives you a link to the location of the data set and the organization that created that data set. So I think it doesn't have to be like, huge detail on where things come from. Just like one sentence up front. Where do you find the data set? Who created it?

Exactly.

Okay, so there's one question about, like, how did you, like, choose a portfolio? Like, how do you build up your portfolio in order, like, step by step. So I think most people, they don't generally build a portfolio. Like, they don't start and go, okay, but I need these, like, 20 different publications in order to demonstrate all my skills. I think that's probably overthinking it. Just start, create one publication and then see what you found. Interesting. If something else, if you see some other data set somewhere, create a publication on it. Just. Yeah, anything you find interesting or anything that looks like relevant to jobs that you're applying for. Go create a publication and don't forget to talk about it on, on social media. If you are, if you're trying to publicize yourself, like, yeah, absolutely, tweet that you've used this, that you've done this workspace, publication. So yes, don't overthink. Like, how do I make the perfect portfolio that demonstrates every skill possible? Just build publications, write publications one at a time and if some of them turn out to be not relevant a few years later, then that's fine. You can always create new publications.

Just to piggyback on. That portfolio shouldn't be static, you should be contributing to them. A lot of times they start with school projects or data camp projects that you've translated from either a course or a project. So you just want to keep evolving it. And the more technically advanced you get, you want to keep adding more examples of the work. You basically want to help yourself so that the next time you are looking for a job, you have something to start with and you're not starting from scratch. There is a question about technical interviews and the expectations for live coding during an interview versus talking through a project and maybe just sort of defending your project or going through the technical aspects of your project. I will say every hiring process is probably a little bit different, but you can anticipate some sort of technical tests. That's usually done asynchronously. It's usually given a certain time frame, a 24 hour, 48 hour time frame to return your analysis. That is usually one of the first stages of a hiring process. So it often happens before you've ever spoken to a human being at that company, whether it's a recruiter or a hiring manager. During the hiring process, anytime you're talking with any team member, you should anticipate there'll be some level of technical questions being asked, both theoretical to test your knowledge on stats or the modeling or whatever it might be relative to that position, but then also maybe specifics where they might ask you some questions after they've reviewed your portfolio. So you really have to be prepared for both of those. And then sometimes you do have live coding challenges where you're given a case study that's relevant to the role that they might ask you to basically whiteboard your response. So there's no single answer. Unfortunately, every, every company handles it a little bit differently, but that's sort of the scope of what could be asked from a technical perspective. I don't know Richie, if you have any other insights to that.

Yeah, I mean it does as, as you said, it varies very widely from company to company. I know Data Camp, we've used like live coding interviews on occasion. They tend to be, they tend to cover like basic skills. So that. Because live coding is a lot of like, well, can I remember how to actually do this? They only tend to ask for like very common skills. So you, if you're doing live coding, you're not going to ask to do something like out of the norm. There was one. I know we're over time. There's one last question that I wanted to answer. So there's a question about just like how do you get that outline menu within the publications? So this is a feature it was just asked, added last Friday in the, in the development version of Data Campus. This is going to be live to everyone later this week I believe. So when you create a workspace publication, it's now going to show you like an outline so you can link to specific sections within the, within the publication. So this should be coming live like later this week. And I don't think we had any other comments within the publication itself. So yeah, that's, that's me done.

Thanks a lot. I think this rounds up the Q and A and the fifth workspace of Live codelong in general. So thank you very much Jen and Richie for your great presentation and thank you all the attendees for your participation and great questions. We will be organizing more of these live code alongs in the future or next one will be next week on Monday. During this session you will learn how to use machine learning with XGBOOST in Workspace. So stay tuned and please let us know if you have any additional feedback regarding this session. I will post a survey link in the chat, but also on Slack along with the recording itself. So your feedback will really make a difference for improving the future events that you hold and to improve workspace for you. If you haven't already, please feel free to join our datacamp Global select community. That's the place where we will keep you updated about new workspace features and events that are happening. So this rounds up the fifth Workspace Live codelong session. I will remind you that the session was recorded and we will share it later on via Slack and email and or social media platforms. With that being said, we will leave you to it and I hope you all have a great rest of your day.

Bye bye.