Understanding AI & Machine Learning for Businesses with their Benefits

 

 

PropertyQuants was founded by a team of Ph.D. ex-quant traders with a track record of successfully deploying automated trading models in global financial markets.
 
They're bringing the quantitative revolution to real estate investing. Property investing today is often based on incomplete and inadequate information. Analysis, forecasts, and decisions are not rigorously produced and are sometimes driven by emotion, untested assumptions, and rules of thumb. Portfolio considerations in real estate are often ignored.
 
PropertyQuants works to improve real estate decision-making. Real estate decision-making should be evidence-based, data-driven, and systematic instead. PropertyQuants are leading the way forward by applying quantitative finance and data science methods to global real estate, helping investors beat the market.

---

Understanding AI & Machine Learning For Businesses With Their Benefits | Nelson Lau & Zain Jaffer

 

I feel it's a duty for everyone who’s working to rather than fight AI, partner with AI because there will be room to co-exist. I don't want to say room to co-exist as if there’s an invasion that is going to happen. Conspiracy is always popular.

 

You think about being able to program and run some beginner data science models, and being able to use Excel when we're ten years into the future. It's a good skill to have. You're not going to get a job just because you know how to do it. That's where we'll get to. Kids are learning this now. I have kids who are of elementary school age. They have after-school-type programs that you can go to. They learn to program a little bit in Python and they learn to run some of these models a little bit. In the future, you have a generation of people who are like, “I know how to use Excel,” in the same way they'll say, “I know how to run around the forest.” It's not going to be the thing that defines their job, but it's the thing that you need to know.

 

It’s a skillset you need to have. You need to be certified or you need to have experience using a computer. I remember growing up, more and more people were worried. There was this whole movement about retraining the workforce and teaching people how to use computers. Something similar is happening here too where the software came in. It might be Excel, Salesforce, or some technology that makes you run more efficiently. It's going to be AI. Understand how AI is used so that you can work around it and you can work with it.

 

It is an education. There's a standardized path here. A lot of people that want to get into this industry struggle in terms of, “Where do I start?” It's not like it’s right in front of you like Excel and you can play around with it. You can buy some courses and learn. People sometimes don't know where to start. What advice do you have for those people who want to make a move into AI? What does the roadmap look like?

 

The other thing is generational. When I was going to do college, I don't think there was such a well-defined AI path that was computer science. Maybe there were a couple of classes in it. Now, most colleges do have a degree in AI and machine learning or something along those lines. There has been a shift, and that's why we have this mixed bag.

 

If you are in this situation where either because you didn't do it in college and you’re a recent graduate or because there wasn't such a thing then, there are a couple of options for you. You can go all the way to try to do a Master's and be online if you have the time and commitment. You can do a generalized course. There are a couple of other courses out there that generally teach machine learning AI. If you're looking for real estate specifically, PropertyQuants offers something. You can come and contact us. We do run a course on how you can apply data science and machine learning to real estate.

 

There are a couple of options there. It depends on how deeply you want to get into it and your budget in terms of time as well, and go from there. The other trend is maybe over time, and this hasn't happened yet, but over the next maybe 10 to 20 years, some of the software might get a little bit more user-friendly. You’re right. With Excel, most people could probably pick it up and figure out how to use it. That hasn't happened yet for a lot of these data science machine learning, but I see some of the things along the way that are going to potentially get us there maybe in the next decade or two. It's fully something where you can pick it up, start playing with it, and figure it out yourself.

 

I struggled, especially when I tried to enter this technical field and tried to figure out where to start. I did endless courses. Stumbling upon your course was the first course I have ever seen that was practical because you're implementing AI in every aspect of the real estate sector. The course is very diverse. We can talk about it later. I struggled initially. I did courses, and some of these courses were over my head. Some of these courses were so mathematical or assumed you had a prerequisite where you could program very well. It felt like some of these courses were made for hardcore engineers who wanted to do more with AI in their hardcore engineering way. What do you think are the prerequisites? What does someone need to know as a basis before they're ready to get technical and understand AI in a hands-on way?

 

We run a course that does get into some depth on this. We don't need people to know programming to start, but we will teach you. People could come in and learn that.

 

 

If you haven't spoken the language before, you need to be able to commit the time to practice and try it out, which is what writing a program is. 

 

 

There's something very powerful about what you said. You can kill two birds with one stone. In order to learn programming and AI, you might as well apply programming to AI. Learning programming is part of the curriculum for learning AI. I found courses like that to be the most helpful. The ones that weren't for the hardcore engineers, but the courses that were built for, “This is a technical course. We're going to teach you the foundations.” It turns out that what you're learning isn’t just core for AI. It's core for appreciating how computers work. It's cool for appreciating how the software works and how computers are overruled, which in itself is mandatory. People need to know this.

 

What is machine learning? We said previously you write a computer program and then do steps 1, 2 and 3. Machine learning is the want to be able to get this outcome. For example, predict the home price. Here’s a dataset, and then telling the computer, “Go figure out how to go from here to here.” That is machine learning and the heart of things like AI.

 

You're right. We run a course that combines these things like programming and then machine learning AI approaches. In terms of prerequisites, people need to be computer literate. There are a lot of people who still have some difficulty with navigating some aspects of that. That's going to be a little bit difficult to get that up and going. Programming takes time. It is a language. In the same way, it takes a little while to learn to speak a new language like German or Spanish if you haven't spoken it before. You need to be able to commit the time to practice, try it out, and try to see things in that language, which is what writing a program is, and then give that a go.

 

Math-wise, most other courses could be similar. You don't need to have a super solid background that you can read and understand the theoretical equations behind it. I don't think that you need to be able to prove why this is the most optimal use of the data. There are some proofs that are like, “This is the most efficient estimator.” There are some proofs that algorithm A is equivalent to algorithm B. You don't need that math, and that's where a lot of the old textbooks would focus on like proof of these things. You don't need that, but I do think that having a high school or equivalent understanding of statistics is useful if you can remember maybe what a hypothesis test is and what is the P-value. After that level, the rest can follow from there. That's our feeling, but we are very reasonable.

 

That's refreshing as someone who comes from the institution of education. You've taught university courses. You've got a PhD. Why is the educational establishment still somewhat outdated and so rigorously focused on the math aspect? Why is that happening?

 

It’s changing now. A lot of these methods that we are calling AI and machine learning are older methods. Some of them were around from the ‘60s. They weren't that popular then because we didn't have the computing power to execute them. They're now coming into popularity. There are more branches and stuff that are probably new. There are a lot of these deep learning approaches that have probably branched off and then gone further. Things like genetic algorithms and all that is probably growing a lot more. I wouldn't say maybe a lot of it is the old method, but that stuff came from academics.

 

You think about how a university is set up. We hire a professor. We hire him because we want to improve the university rankings. He needs to have published a certain number of papers. As he publishes papers, our score goes up. You get these guys who then need to focus on getting this as a job. When I was in the PhD program, I realized there was a shift in my mindset. You go from undergraduate where you're learning some methods. In Master's, maybe you're applying them. In PhD, you're like, "I'm not trying to solve one problem. I'm trying to show in my research that this entire class of problems is potentially solvable by somebody else or that it's not solvable, or that this class problem is the same as that class problem." That's the type of research that happens.

 

As you develop something, maybe you become the one who possesses the syllabus and we end up in this way where it's super mathematical. You teach time series and you focus on all the equations rather than applying, taking a data set, and doing it. That might be changing. The types of courses that we help universities to run tend to be a little bit more light, but then on the other hand, if they are asking industrial guys to help, maybe there is still a gap.

 

It feels like a complete lack of awareness on the academic side. This has always been the disconnect in all systems, the academic system and the practical system where employers are complaining that people coming out of school are grossly under-qualified. The academics also have some elitism going on. This can prevent a lot of people from trying to research and study AI. It feels like a lot of the literature out there is very extreme. On one hand, you've got sensational articles that talk about shallow-level things like how AI is going to cause problems in society and the ethics of AI. On the other extreme, you've got extreme equations and mathematical research. This is what you find when you Google it.

 

 

What's missing was this middle ground and easy path where you take it step-by-step. I can share with you or share with our readers after I've gone through this, not just once but several times. I've done a lot of stops and starts with learning how to code and learning AI. Don’t try to do everything. There are so many areas of AI. We touched them. There are things like computer vision. There is NLP where you're looking at a text and the machine is able to read the texts and the speech transcription. There is image recognition. There are calculations for things.

 

Pick one area, or even better, don't go straight into the algorithms. Start by thinking about something simple like Excel. What can you do in Excel? What are some functions in Excel that are very useful for your day-to-day? One of those might be you can sum some rows of columns together. You can multiply. You can add. You can do all of that. It then might turn into, “How do I find the average? How do I find the mean?”

 

Why don't you start to think about what you can do in Excel and try to replicate that? I found your course and others are a great springboard for this. Get 1 or 2 lines of code and you can calculate the mean. You can calculate the standard deviation. You can calculate some scores. That’s a good way to start. It's not intimidating. It helps to understand some of the intuition of what a mean or a mode is, but you don't have to go into calculus level, differential equations or linear algebra. It's just a few sharp things. It's not the Pareto Principle of 80/20.

 

The few things you can do with your hands on a keyboard can give you a tremendous amount of insights, and how you can plot the most beautiful graph in a few lines of code is mind-blowing. People don't realize that's how easy it is. I'm sure you've seen this with your students. You type a few lines of code and they’re saying, “Wow.” You were able to put that onto the screen and implement it.

 

That’s right. I like this way of thinking about it. Think about a concrete thing that you can do that’s simple. We do try to introduce various things like that. In combination with that, do a project. Build something that has some impact and some interest to you. We focus on that in the course. Pick a data set or run someone else's that's useful for you or your company and present that project. By doing a hands-on, you learn tons of stuff. You learn how it's usable. You’re like, “I need this because I'm going to do that,” or, “I want to show that so I need this.”

 

If your project doesn't involve analyzing images, then don't learn image recognition right away. That makes no sense. The other thing I want to point out is schools are becoming like that. As we hire interns and fresh graduates, I do see a lot of them that have some real-world data projects. Hopefully, that is the trend and hopefully, this continues.

 

To round up, why don't you tell us a bit about PropertyQuants? For people that want to contact you or go to your course, why don't you tell them a little bit about the offerings and how to sign up?

 

You've participated in this before, so thanks for that. We do run an eleven-week course called Applying Data Science and Machine Learning to Real Estate. It is run online. The default is a live session, so you join a meeting with maybe 10 or 20 participants. It’s like attending university calls but online. We have people from all over the world, 40% from the US, about 20% from the UK, and people from all over. We have people who are taking part in Africa and Australia. You get to also join the community and participate. You’re on there as well. You can participate and chat with alumni and current participants. We've had participants from investment funds, pensions, a lot of PropTech guys, appraisers, evaluators, data scientists, and so on.

 

We'll start from the beginning. You don't have to know programming. You can join us and learn Python and machine learning. We'll then go into the domain-specific applications of datasets in real estate like dealing with a property price index, automated valuation, forecasting, how we find similar properties, building market cycle indicators, and geographic information systems so we can analyze the location and spatial data. All of this culminates in a capstone project.

 

 

Machine learning is the want to be able to get the outcome. 

 

 

We've had people listing information and then using that to build automated validation and saying, “Concretely, these are my top ten opportunities to fix and flip and make a profit.” For the people that do forecasting, “These are my top ten buyers to learn.” From a fund perspective, “These are my top ten ZIP codes for office market investment” Those are some ideas or some things you would prioritize in looking at these places. There are tons of content that are applied and relevant. If people want to find out more, they can visit our website at PropertyQuants.com/training or they can email us at Training@PropertyQuants.com.

 

I did the course asynchronously. I didn't tune in live. I know most of the value of the course is live. I was so eager to do your course. You were staggering them. You were doing live sessions and then recording that in the Asian time zone, then the European time zone, and the US time zone. I didn't want to wait for the live US when I wanted to dig in. I benefited a lot from that.

 

I know for the live participants, the courses are also recorded for the duration of that course. They can listen to it. Some of these subjects were so interesting and technical. I had to listen to the videos a few times, but that was beautiful because then you can dissect what's happening and then apply it. You realize, “That wasn't so hard.” That's the beautiful thing about e-learning as well.

 

We do offer that option. I would say the split is about half of the participants doing it live and half of the participants doing it by video. You're right. Even the live participants do benefit from being able to re-watch that. We've taught this quite a number of times now, so we've learned some things about e-learning and how to use this format well.

 

You're right. There are so many advantages over a live class because you can repeat them and you can rewind them. We use breakout rooms. We do exercises. You can ask questions in the community. There are things that happen that you can't do in a regular classroom. We've tried to take full advantage of that. We have tons of support. Although you might choose to do it that way, we have guys who help with the software set up and we have TA sessions.

 

I also think that this course keeps you accountable. It’s not a cheap course. It is an investment. You can probably tell us about what the price is at least. I also found people who do it live, you don't go easy on them. You're like, “Why isn't it working? Show me the screen.” That suddenly changes everything. You can't just passively do this. You have to do it.

 

I felt like a lot of people that do these courses need to be held accountable. We come up with great wish lists when we're outside going for a run or we're on a plane and we write down the top things we want to do in life. One of them is, “I want to learn to code. I want to learn AI.” Do it and sign up for the live course. The benefit of that is being able to share the screen and you being able to troubleshoot issues. There is a setup involved.

 

You mentioned a couple of points there. They're all valuable. Joining the course is a commitment device. If you join either live or by video and we say, “Here's the weekly schedule. Six weeks after the last class, present a project,” that creates a commitment in your mind. You’re like, “I've got to understand this because I want to build a project and present it.” Others can attend the presentation. It’s a commitment in that sense. It’s a commitment in terms of time. There is a financial commitment as well. If you want to find the latest pricing, send us an email at Training@PropertyQuants.com. It does vary depending on which modules you want to take and which attendance option. It’s a little bit easier if we send you the complete information. You can email us at Training@PropertyQuants.com.

 

I'm sure if they mentioned the show, we'll figure out a discount for them of some sort, right?

 

 

Yeah. Mentioned the show. We'll work something out. This is relevant for a lot of the people who are tuning in whether you are in PropTech and you want to apply some of these methods to your business, whether you are in real estate investing, or as you had done where you want to understand what other companies are potentially doing. Getting in-depth into that can be good.

 

Why do people do the course? You do an interview with your students before they join so you can understand them. They can call you and they can find out more. I said to you, “I'm a PropTech VC. I am seeing so many startups who claim to be using AI. I don't know what is real and what isn't. As I'm investing my own money, but I'm investing money from LPs that invested in my VC fund At Blue Field, I need to make sure that I am doing due diligence so I can understand what's real and isn't. Also, I want to open my mind. I want to understand where this industry is going because I want to invest in companies and startups that are going to change the ecosystem of PropTech and generate big returns.” AI is mandatory to get there, at least in many areas. What are the common reasons people do your course? I gave a few there.

 

It turns out that there have been a number of other participants with that same idea. They’re like, “I want to understand the VC space.” We've had a number of participants from that background. We've had some investment funds. This is a popular course for investment funds in real estate as well. They concretely want to build out investment strategies using data-driven approaches. We're seeing a lot more interest in that in Australia and New York. Two Sigma, which is a typical hedge fund, are applying its methods to real estate. There is a lot of interest from that standpoint.

 

We have the funds who send a number of analysts that say, “In our geography and our asset class, can we build a capstone project that does some analysis of where to invest or what the drivers or returns are?” That's one reason. If you're in a PropTech, a number of them have joined because they hired data scientists, but they're not from the real estate field. They’re like, “How do we apply the domain-specific methods to our company?”

 

A number have joined based on that appraisals and evaluations, so a number of forward-thinking appraisers and evaluators say, “I understand if I'm valuing urban high-rises. Maybe there is a big opportunity to scale up the business and differentiate my offering by using these data and methods.” We then have a number of realtors who say, “Maybe you can keep track of a large number of listings and figure out undervalue opportunities or find off-market inventory.”

 

There are property developers that cannot understand the location and some of these factors on the GIS map, and then individuals say, “I understand and it was underscored by this KPMG survey that data science in real estate is the next big wave. Can I future proof my career or get into some of these opportunities, get into prop-tech companies, or get into the forward-thinking real estate investment firms?” It is for all of these various reasons that we've had this mix of participants. There are so many different use cases.

 

It sounds mandatory to me. It is becoming mandatory in schools, at least, where they do teach you some exposure. I can't speak globally, but in many places, they are teaching you how to use computers, maybe a little bit of programming, and maybe a little bit of how AI works. If you're at a university, I highly recommend doing a course. You don't have to jump into a hardcore course. It could be data analytics using Python, which gives you a good entry-level taste. I appreciate you coming on. This is the most exciting topic in my view in PropTech and real estate. It’s great to have an expert like you. Thanks a lot.

 

Thanks so much. I appreciate the opportunity to speak. I will try to catch up again.

---

 

Subscribe to Zain Jaffer: https://bit.ly/2SWhYW5
Follow the PropTech VC Podcast: 
Listen on Apple - https://apple.co/2Izoznu
Listen on Spotify -  https://spoti.fi/2STWDwq
Listen on Google Play - https://bit.ly/2H7s6c0
Follow Zain Jaffer at: 
Twitter: https://twitter.com/zainjaffer
Website: https://zainjaffer.com/
Current Ventures: https://zain-ventures.com/
LinkedIn:  https://www.linkedin.com/in/zainjaffer/

 

Creating the World's First Co-Living Start-Up: https://youtu.be/K2MQNZnfvOA
Innovative Housing that can change the world! https://youtu.be/K92fRT04Qjo
Building an Online E-Commerce Furniture Empire: https:/youtu.be/K92fRT04Qjo

 

About Zain Jaffer:
Zain Jaffer is an accomplished executive, investor, and entrepreneur. He started his first company at the age of 14 and later moved to the US as an immigrant to found Vungle, after securing $25M from tech giants including Google & AOL in 2011. Vungle recently sold for $780M.  
 
His achievements have garnered international recognition and acclaim; he is the recipient of prestigious awards such as “Forbes 30 Under 30,” “Inc. Magazine’s 35 Under 35,” and the “SF Business Times Tech & Innovation Award.” He is regularly featured in major business & tech publications such as The Wall Street Journal, VentureBeat, and TechCrunch.

 

 

Important Links

• PropertyQuants
• PropertyQuants.com/Training
• Training@PropertyQuants.com
• https://Bit.ly/2SWhYW5
• https://Apple.co/2Izoznu - PropTech VC Podcast on Apple
• https://Spoti.fi/2STWDwq - PropTech VC Podcast on Spotify
• https://Bit.ly/2H7s6c0 - PropTech VC Podcast on Google Play
• https://Twitter.com/zainjaffer
• https://ZainJaffer.com/
• https://Zain-ventures.com/
• https://www.LinkedIn.com/in/zainjaffer/