DataRobot

Great time saver for quick exploratory analysis or testing ideas and possible solutions. Not so good for working with very sensitive data or when additional data anonymization is required.

From a credit risk perspective, I think DataRobot is very useful to work on scenario assessments. For example, to identify the worst possible scenario in a default assessment.

Our company had a dedicate scientist ready to answer any question, no matter the degree of complexity and the overall mathematical proficiency.

It is well-suited when you can finally be able to work "easily" when you know how it works. It is not suited when you start to use the tool you need to contract a consultancy, but you have to know how to use it. Also, they have courses you can use, but you have to dedicate the time to study and learn.

I have been using DataRobot since 2020, and they have 4 kinds of people that can help you, and they, for sure, help you. Support helps you with utilizing the tools offered. The key account manager can assist with creating contracts. The money-value data scientists help employees use the tool. Engineers are the deployment experts connecting DataRobot with your company.

Datarobot is a super powerful and useful tool for all data science teams, especially for those teams which are not specialized but where all team members take care of data products from A to Z ( from data foundation to prescriptive analytics) as the tool is democratizing AI. It is very useful as well to gain time, you do not need to spend hours and hours coding as the platform does it for you already. During the training they are providing, you can experience on your own, that regardless of all the manipulation you can do to the models on your own, you will never beat the machine

speed in which you can get model to market ability to mange manage in production

DataRobot is great when you have a structured flat dataset and want to predict either regression or categorization. It is not as well matured in dealing with nonstructured data, images, audio recordings, etc. It is great if you can define your features outside of the software, but it is not possible to make changes to the data once you have uploaded, including performing calculations on the data (e.g. adding two features together).

- When business leaders ask for the fastest possible solution, it is the go-to platform. - When you are not experienced with a specific task and you want to test an array of possible models. - When you are not a fully devoted Data Scientist, and therefore cannot spend too much time on fine-tuning model hyperparameters.

If there is some experience with data and especially if there are enough resources and time available for users to invest in the projects. It is unsuitable as another side-project that can only have minimal resources. Ideally, there are already some ideas for concrete ai cases that could use the company.

Appropriate When we have a very straightforward data science problem, which needs a scalable solution, then it is a suitable solution. Suitable for time-aware models, DataRobot can created many features. Less Appropriate When the data science problem is challenging and requires lots of fine-tuning, the DataRobot is not a good solution.

We used for some of our business problems to get the most appropriate features, producing the solutions and ML scenarios

It is especially best suited for companies with many AI use cases due to its rapid prototyping. It is also beneficial when the data science team is not big since it can multiply the productivity of the team, allowing each data scientist to work on more than one project at the same time, leveraging the platform speed. Also good when the data scientists are not experts and can benefit from the guardrails that the platform offers.

DataRobot is well suited when the business has really defined its use case well and you have internal SMNE's/resources to support the use cases. In addition, you need people with some data science and/or analytics knowledge. It is less suitable when the above scenarios don't exist.

Predicting customer churn of our top 30% of our referring doctors and the effects of them doing so Predicting IT service tickets using keyword analysis, so as to save time from having to manually select different categories and sub-categories, eliminating delays for tickets to be picked up by service desk agents.

Having DataRobot is like having 100 Data Scientists quietly humming away behind my screen. And, the great thing is these 100 Data Scientists seldom argue and actually agree !! "They" are very efficient too:) Automated Machine Learning is here to stay - it's like outsourcing analytical knowledge and insourcing business knowledge - the best of both worlds for a humble school system with over 43000 students.

It is good for bfsi industry to use this platform to find details regarding customer needs and their onboarding journeys. Also useful to find out the problem areas from where the organization is not getting the required revenues that could be cross-sold, digital sales, partner sales .etc. This platform may not be useful for tech-related statistics

It's very superb for business intelligence, and making business decisions. It has intuitive UI, and it's excellent with data consistency.

for scenarios where plenty of data is available and a decision based on data is to be made. DataRobot is less suitable for unsupervised learning and scenarios where enough data is not available.

The DataRobot platform can create numeric predictions, binary and multi-class predictions, and time-series forecasts. Like any ML problem, DataRobot works best when your ML problem is well-framed and the dataset is formatted appropriately. DataRobot has built-in guardrails to prevent novice or inexperienced ML users from creating nonsensical models. The platform does a great job of presenting visibility into how the model determines a prediction. The explainability embedded in the platform makes it easy for non-data scientists to understand the models that were built.

Datarobot is one of the best AutoML tools out there with many SOR models. It has helped our organization in realizing the potential of AI and AL. It is really easy to use and understand even for a budding ML engineer or data scientist. They also provide the best support and continuous improvement to the tool.

DataRobot gave us models we could work on within a matter of days completely solving the pure machine learning part of the equation. People in upper management still have a hard time grasping how because of your tool we no longer need help with the AI part of things. I keep referring to your tool as a beast that needs to be fed period.

It is well suited for analysts like myself who are familiar with predictive modeling and understand some of the behind-the-scenes aspects but aren't to the point where they could program a predictive model by themselves from scratch. I think I've had the most success with it as an analytical accelerator. If you have no modeling experience or aren't familiar with some of the concepts, it will still work, but it could be a bit overwhelming/you might lack the knowledge to make intelligent business decisions with the results. On the opposite end, it can definitely help seasoned modelers who are able to do things from scratch themselves, but it can provide a hindrance if there are specific tweaks you want to make that aren't available in data robots or are two hard to decipher from the back end.

For companies that do not have specific in-house data scientists, the DR platform and support system is exactly what is needed to get started and integrate modeling as a supporting capability

This is a good way to take open-source tools and add in professional support to take some of the guesswork out of which libraries to use. The ability to apply multiple tools in parallel and rank the results is particularly effective for us.

DataRobot is well-suited for exploratory analysis that requires [the] reduction [of] thousands of variables to a handful of most impactful for predicting the target. DataRobot excels at serving complex model building where accuracy trumps interpretability, as well as offering the ability to manually build more interpretable, less complex solutions.