Indoor navigation and machine learning combination both for helping users to find the most suitable stores and for helping stores to advertise their products. This is my first article for the last few months and I am incredibly happy to get back. During this time my life turned in a different direction and now I’m doing something completely different from what I did half a year ago. But still, I remained faithful to neural networks, machine learning, and technology in general. Now I am developing a mobile SDK and Android application at Navigine team. A few words about the Navigine — comprehensive indoor positioning platform (I did not come up with anything, just took a phrase from our website). I think it should have become clear to the reader that today we will talk about indoor positioning and navigation. If you read at least some of my articles, it will become obvious to you that I am a big fan of machine learning and augmented reality and therefore today’s article will certainly not only be about indoor navigation but about its use in conjunction with machine learning. Well, let’s try to understand all the opportunities that these two technologies provide for us and maybe some of you will be able to create something incredible soon. Or maybe it will be me! Implementing Indoor Positioning First, let’s deal with indoor navigation, and then move on to dessert, that is, I wanted to say to machine learning and predictions. In the field of indoor positioning and navigation, there is not only one single solution, already on the first page of Google search you can find some good options, but since I am well acquainted with our solution, I will talk based on it. First things first, you need to create the digital map of your location, in the case of this article, this will be the map of the shopping mall. Now you need to add the Bluetooth beacons to this location and link them in the app to the places where you deployed them. By the way, the more beacons you add, the better the quality of positioning and navigation you will get. As for beacons, I will leave you the choice of their producer, dear reader. Do not forget to add zones, barriers, draw routes and mark all the venues, so the user will be able to find the place he needs faster. And the next step is the most interesting part for me — using Navigine SDK in your iOS and Android applications. You can use Navigine SDK to add indoor navigation, push notifications and tracking functions. Let’s take a deeper look at each of these opportunities. — it is obvious that users will use it to build optimal routes to the destination points and to find their positions. But if you look at the title, you will understand that we can get some more interesting information from this point. Indoor navigation —Push notifications are a separate very extensive topic, so I will cover this in another article. Push notifications — this will allow you to see the full route of the user at every second when he uses the app. As you probably already guessed, on this functionality we will focus on the next part of this article. Tracking functions It becomes obvious what kind of data we can get using , so let’s start moving on to the most interesting part of this story. Further, the story will go on how we can use this data, what things and events we can predict and what interesting functionality we can add to the Shopping Mall application. Navigine SDK Machine Learning and Data Science part In this section, let’s look at two cases when the user does not need to register in your application and you do not receive any data about him. Well, and the second case, when you know his age, gender and some other personal data that he had to enter when he registered. Of course, the second case gives us many more opportunities in predicting and analyzing data but even without these data, you can achieve any results. Without user data In this case, we will have to get out on the basis of the data that we have, but even here we can achieve great results. Even if we don’t have user information, we still know that repeated searches for routes are made from the same device and therefore we understand which routes belong to a specific user. I will give some examples of how you can use the data, but of course, no code will be given, as well as explanations of how to implement this, I will leave it to you. Knowing the routes each user builds, we can group them according to these routes. We can already use these data in different ways, for example, to analyze at what time each group goes to shopping centers and thereby we can offer them certain services. Also based on this data, we can determine the interest of new users by their first few searches. Or for example, to notice different trends, as one group of people after a certain time begins to be interested in other things and use this knowledge in the right direction. These are probably the most obvious options that come to mind, but with their help, you can achieve very large and interesting results. So let’s move on to the way when we know at least some user data and see what obvious solutions we can find. With having some user data Imagine the same points as in the previous paragraph, but just add there that we still know the gender of the user, his age and some other personal data, such as the address of residence or something like that. Now we will be able to do clustering not only by interests but also by many other parameters, for example, by sex, by age, by place of residence and notice the exceptional interests of each group. And of course the most important thing, now when registering a new user, we can immediately recognize his interests and predict which stores may be more interested in him, which will allow us to advertise the right offers and. This data will be useful not only for the users themselves, but also for the owners of shopping malls, they will understand which stores are more interesting for users, this will allow opening more similar shops, also knowing these they can offer the correct rental value, and many other interesting solutions and useful information can be fished out by holding a small brainstorm. Conclusion Summing up all the above, I want to say that indoor navigation can be used in very interesting places and based on its data, you can analyze a lot of interesting and most importantly valuable information. Technologies allow us to improve our lives and reduce the number of tasks that we have to think about, freeing up much more time so that we can do more useful things. No matter how you feel about it, the time for great technological solutions has already come and every day it improves our life, even if we do not notice most of it. Therefore, I wish you to advance in the study of technology and science as far as possible and I hope someone will benefit from this article. Previously published at https://towardsdatascience.com/indoor-positioning-and-predicting-the-most-suitable-boutique-for-customers-in-shopping-malls-8149b3097b57
