A machine learning prediction of our vaccination speed
WHEN WILL I GET MY COVID-19 VACCINE?
We figured this question is on everyone's mind these days. The fact that 140.000 Belgians were waiting in line to be put on the reserve list of Covid-19-vaccines (and not Tomorrowland tickets, as is usually the case around this time of year) made this clear already. Moreover, at RoboRana, we are looking forward to getting back to normal and being able to work at the office together with our colleagues. Therefore, we felt that it was time to offer smart insights into the vaccination data publicly available with an intelligent dashboard application. Of course, we couldn’t wait to bring our application to you, so you can look forward to a bright future, together with us.
We’ve put our end-to-end intelligent automation expertise to work and developed a website where you can check when a certain percentage of the Belgian population will be vaccinated, based on available government-issued data. For instance, if you’re wondering how much of the Belgian population will be vaccinated by this summer at this pace, you can check it via our vaccination prediction tool.
So, are you ready to see our predictions? Check it out:
Wondering how we’ve made this possible? Let’s dive into this wonderful example of powerful intelligent automation, in which we applied the combination of automation tooling (RPA) with artificial intelligence.
RPA & AI: A POWERFUL COMBINATION
By combining Robotics Process Automation (RPA) and Artificial Intelligence (AI), we were able to create an autonomous flow that results in the visualization you can find above. The first step of the process was putting our RPA bots to work. As we needed to be able to make predictions, we needed data to feed our AI algorithm. To get this data, our bots scour the internet for relevant official vaccination data, provided by our government, which are then sent to an Excel file by the RPA bots. In addition, the bots keep adding new data to the existing rows every day, so the model keeps training.
Talking about models or algorithms, let's look into the Artificial Intelligence side. We built this with a Python program that uses the Belgian vaccination data to make predictions about the future through machine learning (ML). Machine learning is an AI field that focuses on algorithms that can find certain patterns in data sets. As there are hundreds of ML algorithms that try to do the same thing, it is quite a difficult task to pick the best algorithm. As a result, there is a lot of trial and error involved in finding the best ML algorithm. As a result, it is extremely time-consuming to do this all by yourself.
Time-consuming tasks? Sounds like something we should be able to fix. We wouldn't be hyperautomation experts if we didn't find a way to speed up this process. Therefore, we used Microsoft Azure, a collection of web services, to point us to the best model, that was chosen from hundreds of different models that are trained simultaneously. The models were compared on several properties, including accuracy and drift.
Microsoft Azure made it possible to find the best algorithm in an hour instead of days. To continue our technical explanation, our application uses a deep learning algorithm to find patterns in the data and make the right prediction. This means that the program creates a kind of "rudimentary brain" that can find connections in the data. Once the model was established, we published it to an API so we could also share it publicly and privately.
As a final step, we have chosen to create a website on which everyone can check when a certain percentage of the population will be vaccinated. As we are writing this blog post, the Belgian population will be fully vaccinated by September (based on current data and vaccination progress). Can our government beat this number? Keep checking our vaccination prediction website, you might be surprised!
MINIMUM VIABLE PRODUCT
The best thing about this application is that it's still a minimum viable product (MVP) in which we combined our expertise about process automation and AI. We can expand this website and for example, add other countries, or make it more detailed as to which parts of Belgium get vaccinated more quickly. As long as there is data, we can keep improving. The beauty of artificial intelligence, right?
Let us know which feature you would like to see!