Find a Data Science Project With These 3 Techniques

Key Takeaways

[Music] okay so many people i talk to struggle to come up with data science projects and to me identifying problems and digging into them is one of the most important parts of becoming a data scientist to begin with finding interesting projects is a skill and is something that you can improve on in this video i'll highlight the three techniques that i used to discover projects to work on in my own personal life i get asked to recommend projects to do all the time this is something that quite frankly i refuse to do i'll recommend the types of projects like i've done in the video i've linked above but i'll never tell you what data to use or what topic it should be on why is this if i tell you exactly what to do you're missing out on one of the most important parts of the data science process understanding how to identify problems and solve them is an integral skill in this profession you learn this skill by finding and building projects in the real world plus if i told one person a bunch of other people would also do that project so it would no longer really be original right now i'm going to show you how to fish rather than just give you the ideas this will serve you far better in the long run it'll also serve me far better by not having to say no to so many people asking for project ideas okay let's start with technique one think about the problems you face on a day-to-day basis would data help make your life or someone else's life easier or more understandable we use multiple apps every day we have an exorbitant amount of data that's collected on us don't we have the right to analyze it i think that it's more important to start with problems and then drill down to projects the more specific a problem the easier it is to create a project around it when i was in grad school i was dead set on starting a company i would do a thought exercise every day i'd write down 10 business ideas that i could think of they didn't have to be good and trust me some of them were bad but i had to get 10 on paper i found it extremely difficult to do this when i was thinking of big picture ideas but it got easier for me to find 10 when i focused on an industry or a specific niche for example instead of just thinking of 10 things i would think of 10 different ways to improve airline flights or i think of 10 different ways to smash the like button and you know subscribe and turn on notifications of course you can do this exercise too get into the habit of identifying problems and projects every day try to think of 10 and just get something on paper start with something that you're interested with maybe 10 ways data could make golf more enjoyable or 10 ways to improve your valorem performance with data or 10 ways that you could use data to create an anime series it's way easier to come up with these ideas if they're close to your heart for example in one of my project review videos some realized they loved exploring new music but they also found that they wasted a lot of time previewing songs that they didn't like they use their spotify data to predict which new songs that they might like they made this recommender system very specific to them this immediately saved this person time more recently i've been exploring different ways to understand you better and by you i mean the youtube audience i started collecting data from my comment section i wanted to see if there were trends in what people were saying or if recurring questions kept popping up this analysis can directly help me to improve the content that i produce so long story short focus on finding problems and also practice coming up with problems using that exercise okay i think that's that's enough for technique one might uh might have beat it to death a little bit let's move on to the second technique this is looking at data and finding something that appeals to you this one might be a little easier for most people but i don't generally find it to be as rewarding the approach here is to comb through data on kaggle on google data sets or any other data hosting websites and to find a data set that really piques your interest from there you should spend a lot of time exploring that data again i find these slightly more difficult because i get far more motivation in having an end goal of an analysis exploratory data analysis can in theory be limitless if you have a large enough data set so i recommend doing some light exploration and then establishing some really clear questions that you want to ask of the data with the knowledge that you have again i recommend finding data sets that are less commonly used for this because no one will be impressed of your analysis of the titanic data set or the real estate data set or the number image data set unless you do something completely outrageous with it projects based on existing data will almost never be as interesting to employers unless it's fundamentally different than the other projects and notebooks that are out there on the subject one way to differentiate with these is to compete with them on kaggle even if you do a similar analysis to other people if you perform well in a kaggle competition you'll be able to show employers that you can think critically about a problem and you produce good results in order to do well in these you generally have to think outside the box so you're still showcasing that ability admittedly this can also be very frustrating as there are a lot of really smart people on kaggle regardless i don't think anything great comes out of this field with anything less than you know a little difficulty okay on to the final technique it is getting feedback from communities something that i've started to love more recently is getting involved with various online groups most of these communities have their own questions and problems whether it's finding golf courses that are open during a pandemic or which automl tool makes the most sense for your use case there are a ton of discussions on reddit on youtube on linkedin on facebook and you know these other community building websites these are literally filled with people who have problems just keep your eyes open and you might see a few of these that can be solved using your skills an incredibly meta project is doing a project where you analyze the questions that are coming in from one of these communities you're a part of perhaps you could create a chatbot for one of these set communities perhaps that community could be the 66 days of data all the projects that you do should be creating value for someone and again this someone can be yourself creating value for communities you're part of is almost always welcome and can get your work shared this is how my data science project from scratch series came about as well which i've linked again above and below i was looking for a project that would be useful to my subscribers hopefully helping them to better understand the job market and the job postings that are out there if you're well known in a community you can also ask questions about what problems they're currently facing you don't just have to be a fly on the wall in these scenarios i find that this approach could potentially even reveal some business opportunity and there is a very big difference asking communities for problems that you're looking to solve and asking me for a data science project when you're asking a community you're eliciting feedback for something that isn't you know specifically within the data science domain you're finding the data science project within what these people are saying whereas if you're asking me you're just saying you know you're kind of just being lazy right throughout this whole thing notice that i didn't get caught up in fancy algorithms or anything like that we're focusing on solving real problems with tangible outcomes this is what employers this is what pretty much anyone looking to vet your data science ability is going to look at okay that's all i got i hope these techniques help you to come up with some killer project ideas if you share them online make sure to tag me on linkedin or twitter i almost always dish out the likes and the shares assuming i see them until next time thank you so much for watching and good luck on your data science journey you might see some videos popping up over here if you enjoyed this video definitely consider checking one of those out next