Darwin Review

Helps us transform data into knowledge faster by selecting the best algorithms for us


What is our primary use case?

The primary use case is to predict the default on payments by clients.

How has it helped my organization?

We have a lot of data sets with different kinds of mistakes. We're using Darwin to help us to fill in the blank spaces. It helps us to solve that problem and clean the data when we have blank spaces in it.

We are increasing our number of operational algorithms. We are developing algorithms to predict data for the next month, including forecasting our income using macroeconomic variables. About 80 percent of our models are operational.

We are able to develop models faster. We are a small team; there are three of us involved in business intelligence. We develop models faster because, instead taking three weeks to test a lot of algorithms to select the best one, we just upload information to Darwin and let the software help us define which is the better algorithm for that data set. We developed a model for a customer and we are on track to have a new version of that model in two months. We can do that faster because of Darwin.

Darwin enables our company to tackle more complex problems by making data science more approachable and operational. While we need people who understand what's happening in the business, the solution helps us to use our time better. We need to be more creative to have more variables that we can use for making models. We are working on creating bigger and better data sets, to increase the quality of our data sets, and on the automation of the model and the result of the model.

It has increased both efficiency and productivity. It's helping us to improve our data sets and add new variables, instead of proving and testing a model for multiple weeks. Our productivity is increasing because we are making better decisions. We are making a change in the company, becoming a data-driven company and really using our data better. For example, with our recent development for the sales team, there has been a 20 to 25 percent improvement in efficiency because we developed a solution that helps the sales team to identify when a client has enough money.

Our main goal is to transform data into knowledge and Darwin is definitely helping us to do that faster. An example is the issue of loan defaults. We have a lot of data on loans that we have already made to clients and, on some of those loans, the client defaulted in payment. We took that data and converted it into a model that helps us to predict which clients might default. We have started to cancel those loans and recover the money. Without that data, we would have a higher rate of payment default.

What is most valuable?

The thing that I find most valuable is the ability to clean the data. 

In addition, it helps us to create a model. Instead of trying things one-at-a-time, Darwin helps us to improve models and select the best one.

What needs improvement?

Our main data repository is on AWS. The trouble we are having is that we have to download the data from our repository to bring it into Darwin. It would be great if there was an API to connect our repository to Darwin. It would provide great automation because right now it takes time to download the information and then upload it to Darwin.

Another area for improvement would be if the user interface could have non-supervised models. That would be great. Right now you can only work with supervised models.

Finally, I would recommend that they work on improving the account functionality because we have had some difficulty in that area, in terms of logging in.

For how long have I used the solution?

We have been using Darwin for almost eight months.

What do I think about the stability of the solution?

In the first version, I think they got some feedback, but the stability has improved with the new version. We did have some problems when we tried to upload some data but it's getting better.

What do I think about the scalability of the solution?

The solution has scalability. The only thing that I would recommend is that it provide folders because we are using a lot of data sets. As we get more and more data sets and models, it gets a little bit complicated to find the model or the data set that you are looking for. Regarding scalability, that could be an issue in the future.

We have five people using Darwin. Three are in IT and two are members of the management team.

We don't have plans to increase usage of Darwin because our company is not that big. You need people with certain kinds of knowledge to work with Darwin. But currently, we are in the optimal position in terms of number of accounts.

How are customer service and technical support?

Technical support for Darwin is good. We have had some trouble with our accounts in terms of being able to log in, but they responded to the issue very fast.

How was the initial setup?

In the beginning, it seemed a little complex. We needed an introduction and some sessions to help us to understand what was happening. Maybe it would be easier if we just didn't ask any questions and just uploaded and ran the model. But we had a lot of questions about the results and how to compare them. That was the complex part. But the main reason we had a lot of questions was because our team, in general, has a lot of questions and wants to know more about everything.

It took us about six to eight weeks to deploy Darwin.

We already had a group working on machine learning, so that made it easier. Before Darwin, we worked with Titan and RStudio. The complex part was to take models we had already deployed in our pipeline and make them in Darwin. Our strategy was to first prove the models that we had already made, and then to start on new models. 

Now, we are aiming to have a model that will help us make the decision about whether we should make a loan or not to a given client.

What about the implementation team?

We developed things with the help of SparkCognition. We have a really good team. We understand how AI works. We didn't require any consultants.

What was our ROI?

It will take a little bit more time for us to create more models and convert more data into knowledge. I'm sure that we will have the return of investment that we aim to have. It will take some time.

Which other solutions did I evaluate?

In my studies, I work with a solution called BigML.

What other advice do I have?

You need to have good data sets to get good results. Before Darwin, you need to work on your data sets to have the correct data sets to make the correct models. Darwin is a solid solution, but the main advice that I have is that if you don't have the data, you can get Darwin but you're not going to get the results you want.

The biggest lesson I have learned from using Darwin is that it makes things faster. We can test faster, not just one at a time. We speak with the team at SparkCognition and they help us to improve our ideas around the use cases that we can apply. That is another important lesson.

The biggest problem for us is data sets because, sometimes, they don't pass in relation to Darwin. It's not a problem on Darwin's side, it's a big problem for us because we have a lot of unstructured data and we are working with other solutions, not Darwin, to have the data ready for algorithms.

For Darwin, as a solution, you need people who understand the business and who understand how to improve the organization with the results of the models.

Which deployment model are you using for this solution?

Private Cloud
**Disclosure: IT Central Station contacted the reviewer to collect the review and to validate authenticity. The reviewer was referred by the vendor, but the review is not subject to editing or approval by the vendor.
Add a Comment
Guest