The Real Data Science Experience
February 21, 2018
The Real Data Science Experience
Score 10 out of 10
Vetted Review
Verified User
Overall Satisfaction with IBM Data Science Experience (DSx)
DSx is used for people that want to collaborate on projects concerning Machine Learning and Artificial Intelligence. It was used to create a recommender system on an iPython notebook, a classification solution and currently a genetic algorithm implementation. Overall, the advantages it provides is a stable platform, where users can run online a solution, save the results and collaborate, which seems to be very useful for our organization. It is mostly used by the department of Analytics but the results are viewed and used by managers of all departments. It particularly addresses problems that have to do with the exploding amounts of data and monitoring performance as the user can save and control their data and results.
- DSx has a very straightforward UI, that is simple and easy to use even by users without prior relevant experience.
- DSx has cloud Implementation enabling data scientists and analysts to work on a project collaboratively and store all the data and results they produce on the cloud.
- DSx uses open source solution and brings together the state-of-the-art tools for data science: python and R, on a single platform.
- Another very important advantage is the learning aspect of the platform, as it guides the user with tutorials and good documentation, making it simple to use by non experienced users.
- The kernel of the platform has been quite unstable from time to time causing problems to running code and results.
- The collaborators of a project do not have the option to run code simultaneously on the platform making it difficult to actively achieve collaboration.
- While R and python are the 2 major analytics tools, there are many more that exist and could be implemented to achieve improvements in results and to attract more users with different analytical and software development backgrounds.
- After using the DSx we produced predictions that were approximately 1% more accurate which may seem low but is extremely hard in machine learning.
- We improved our response rate by 20% as the projects were more organized.
The IBM Data Science Experience enables data scientists to collaborate through projects, to which they can add notebooks, data, data connections, and other users they want to collaborate with. In Jupyter notebooks they can use Python, R, or Scala, when needed with Apache Spark, to analyze data from diverse data sources and data sets, and can share the outcome with stakeholders or the public via URL or by exporting the Notebook .ipynb file and publishing it on GitHub. Additionally, RStudio is included. Overall, DSx combines all the popular tools in a single platform, and that is the main reason we selected it over one single tool.