We use different data science platforms for customer-specific projects. Whatever is being requested by, or is required by the customer, we learn it. Python is one of the technologies that we have a lot of experience with, and it is part of Anaconda.
Our primary use case is analytics. We use Anaconda to build models that predict the probability of an event, or it can be used for classification purposes. There are various uses for this tool.
One of the things that we do is subrogation and I can explain by using the example of a car accident. When an accident happens, you take your car to your insurance company and give them details about what happened. Also, the advisor at a service center will write down relevant information and supply it to the insurance company as well. At this point, the insurance company reimburses expenses for all of the damages that you have incurred. At the same time, they would like to find out if there is any fault that can be attributed to another person. If so, then they want to know whether it is possible to make any kind of recovery from that person or their insurance company.
With thousands of these claims coming into the insurance companies, it is very difficult for somebody to read all of the information and decide whether there is a potential for recovery or not. This is where our application comes into effect. We read all of the data into our software, which is built with Python using Anaconda, and try to gain an understanding of each and every case. This includes many details, even claim history, and we try to assess what the chances are of recovery or what the chances are of subrogation in each case.
This is just an example from one of our several clients. Each customer has different requirements and we customize a solution based on their needs.
The most valuable feature is the set of libraries that are used to support the functionality that we require. We use different libraries for finance and numbers, and we use the scikit-learn library for machine learning. A few of these libraries are very helpful and there is a very long list of them.
I think that the framework can be improved to make it easier for people to discover and use things on their own.
They need a better interface because currently, we have to do everything through coding. It would be nice to have a simple description of what each library is used for and how to use it.
I would like to see additional libraries included to support computer vision and natural language processing. The framework gives us the ability to create them, but having more in place would mean that we would need to do less coding.
Stability is not something that we really consider for this solution. When we are using Anaconda, we have to develop most of the things from scratch. It's a framework, and it is one of the tools that we use so that we do not have to think about dependencies. When I have Anaconda in my environment, I do not have to think about any prerequisites that may be required.
We have not been in contact with technical support to this point.
The initial setup is straightforward and not too difficult.
The length of time required for deployment changes after the first time. If somebody has to build everything then it takes longer. However, once all of the libraries are built, it takes one person perhaps three hours to deploy into production if it is done without interruption.
We have our own team for deploying this solution.
This is a great tool to work with, even if you are starting your career in analytics or another stream like data engineering or data science. This is a tool for everyone because you don't need to think about many things, such as what needs to be installed.
I would rate this solution an eight out of ten.