Hey everyone, welcome to another episode of Data and AI with Mukundan. So in this episode, I want to talk about dynamic topic modeling, and how I can simplify this for the everyday person. So dynamic topic modeling, in a nutshell, it helps you understand trends in your data and make smarter decisions. 

But if I were to give you like a more deeper explanation, it's a way to track your changes in the topics or themes over time. We look at topic modeling a lot in data science, and I think this is a topic which is like a more niche. In the field of topic modeling, it's just not talked about as often.

So I just wanted to cover this and tell you why it can be, you know, a game changer for businesses. So, dynamic topic modeling is used to turn text data into actionable business strategies. What is dynamic topic modeling? It's like a machine learning model that tracks the evolution of themes in text data over time.

Think of it as finding out what people care about year by year. It helps businesses see patterns in data like customer reviews or social media posts. And why you should care is, basically, it helps you stay updated.

You can get to know what your customers want today, what they don't want, helps you predict trends, like I said. You can stay ahead of your competitors because you know what your customers are wanting, and you can see how that's evolving over time. And it helps you save a lot of time.

Basically, you are able to understand huge amounts of feedback pretty quickly. It makes better decisions as well. So basically, you're using data to guide your actions, right? So you're using the insights that you gathered from this dynamic topic model, and then you're making smarter decisions because of it.

All that's great, but you'd be wondering, how does this work? Basically, it just looks at text data. So any kind of reviews, reports. So especially for businesses, if you want to look at how does this work, you would look at text data, like reviews, customer reviews, or, you know, the reports about something over time, maybe something important in your field, and you want to track how that report changes over time for that topic that you're interested in.

And it groups words into topics like price or quality. So it will take like a large piece of text, and it will make it into a smaller subset of just one word, like a price or quality, and helps you track how these topics change over time, like from 2021, 2024. So you can see how something has evolved in the last three years.

So for example, if you have a product that you are selling, maybe like, let's say you're selling like a digital product that your company makes, and you want to track reviews of that product over time. What are people saying about this? Let's say, let's say you have a productivity app, right? That's the digital product that you're selling, and customers review it in 2021. So they're talking about, let's say, the price and the quality.

But in the next three years, what they're talking about is, you know, a special feature of the product, and less about the price and quality, because they may have been like really happy with the price or whatever, right? So that's something that a topic model can help you track over time. And I don't know if you've seen this, but Amazon has these customer reviews, and it has like, you know, these reviews, which are categorized by different topics. So if you look at any customer review, let's say you take any product on Amazon, and you go to the review section, you'll see like these reviews categorized.

Some of them will be talking about the price, some of them will be talking about the durability of the product. So things like that, that is nothing but a topic model. If you, yeah, if you pay close attention to that, you will actually see that.

So yeah, that was like one real life example I just spoke about. But if you look at, if you want to look at more examples, let's say, let's take marketing, for example, you can adjust your ads in marketing to match what the people care about, right? Let's take these customer reviews. For example, you understood your customers are speaking more about a specific feature more than your price and quality.

So you can adjust your ad to match that, you know, customer need. Customer is talking about the specific feature, maybe market to that feature, right? And so it helps in that. And it helps improve features based on customer feedback on a product level.

So like I said, like you already know what's working, what's not because your customers are telling you that. So you can improve your features. Like if they tell you like a certain feature is lacking something, you can use that as like a fuel for going to the next step.

It helps you spot trends. It helps you spot what's becoming popular, like sustainability of like maybe of recent last three, four years that I've been paying attention. I'm seeing a lot of increase in sustainability-based products, right? And so that's, I mean, maybe if you're selling a physical product, sustainability becomes a big thing.

So your customers maybe want, maybe talking about that, like is this like, you know, environment-friendly or something, right? You can utilize reviews like that and make your product better. Now, if you look at what your employees are talking about, so I even spoke about earlier, like, you know, how are you looking at your employees? So like it can be used to improve your human resource strategy to maybe help predict employee churn and even reduce it for that matter. You can see what are the trends in employee behavior.

You can see what the team is telling about you, like your customer, your employee team is telling about you. So that kind of thing over time can be tracked. Another more simpler example.

So let's say in 2021, customers love the design. 2022, they wanted better battery life. And 2023, they care mostly about the price.

So dynamic topic modeling helps you show this kind of a trend so that you can adapt your product to your customers' needs. So how will you go about doing a dynamic topic model, right? So basically, you will start with collecting your text data. So for example, collect your reviews, collect your surveys, and use tools like Python to analyze it.

You can get clear visuals and reports to guide your decisions. I've been using Python specifically to do dynamic topic models, but I'll try to find other tools out there in the market which already give you like an inbuilt, basically, they already give you a product which does dynamic topic model for you, right, a DTM for you. And it helps you get clear visuals and reports which help guide your business decisions.

And that's really the aim. You can already see how this can make your business better. It helps you understand people better.

That's why it matters. It saves time and makes your work easier, gives you a big advantage over others in your field. So that's a pretty significant advantage.

And again, it helps you stay ahead of your competitors. You can keep fixing things on what's working, what's not to help you turn over a big profit as well in the process. So there's a lot of things happening.

With all that being said, I'll start using dynamic topic modeling today. Make smarter decisions with your data. And if you need help, just feel free to email me at mukundansankar.substack.com. And I'll write that email in the show notes.

So just feel free to reach out to me. I would say just like play around. And if you're not a data expert, reach out to one.

And if I'm not available, because I'm still a part of one. But anyway, have fun with this tool, because it is going to change your business and your life. That's all for this episode.

I will see you again in the next one. And hit that subscribe button. So you don't miss another episode of Data and AI with Mukundan.

Thank you. And I will see you in the next one.