Reasons why data science projects are not always successful – Part 1

[This article was first published on R-Bloggers – eoda GmbH, and kindly contributed to R-bloggers]. (You can report issue about the content on this page here)
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.

Data science is one of the most wide-ranging disciplines of the 21st century. Data scientists use a wide variety of methods and tools to generate more knowledge from data and its analysis. Especially in times like today, data and the insights we can draw from them are becoming increasingly important. Almost every business process generates and uses data – in fact, almost every one, if you leave out the distinction between digital and analog.

People often think: the use case is defined, the hoped-for business value is clear – but it still happens that not all data science projects create the promised added value. What this may be due to, which factors need to be considered and how, for example, a tool must be set up to help in this case, will be discussed in the following.

1. The proper team

No project can bring company-wide success with just one single user group. Business experts cannot develop analysis scripts, algorithms, or a platform that makes these scripts productive. Software developers can develop the platform but have no influence on the technical infrastructure. Data engineers know the requirements of a high-performance infrastructure but have to rely on the others for development. Data scientists develop the analyses but are dependent on the input of the others for the business context and the infrastructure. Finally, users have to finish the job.

Once the use case has been found, all participants MUST work together! This starts with the conception, continues in the development and ends in a permanent evaluated and productive use of the solution. One solution is the DataOps approach. Here, the basic principle of the DevOps model is used and extended by further tools, methods and the above mentioned actors. The aim is to extend the process to A by the above mentioned groups with the aim of using, optimizing and transforming the (further) development of data analyses and their results into data products. In short, successful projects require the appropriate know-how from business expertise, data science, infrastructure and all those who will work directly with the solution. Does every organisation need a whole football team for this reason? Here too: No! Business context and users are available in every organization. Expertise in the areas of data science, infrastructure and the implementation of a suitable platform can be provided by a strong partner who works closely with the organisation and the people involved. This creates increased acceptance of the projects and their results.

 

2. Acceptance – the user must be in focus

It often happens with digital services that they have to struggle with a lack of acceptance. No solution is promising if the users have more problems than benefits. This can be due to an overload of information, cumbersome dashboards or simply unnecessarily complicated menu navigation.

This must be avoided:

  • Cumbersome usage, which rather hinders the daily business – can be solved with clearly arranged dashboards / views, which can also be individualized.
  • Poor performance of the solution – nobody likes to wait 7 minutes for a report to be generated.
  • Lack of traceability of the results – how did they come about? Focus on transparency instead of black box concepts.
  • Missing or incomprehensible presentation of results – nobody wants charts that are not understood.

Unnecessary switching between different solutions for different use cases – drag a table from solution X and then create a chart using Excel? No thanks! The work must be done without media discontinuity. The good news: The market has reacted and developed appropriate platforms. A data science platform such as YUNA is taking effect due to the modular structure of YUNA. Dashboards can be set up specifically so that only the desired information can be displayed and used. The results can be filtered in a configurable way and each data point of a chart can be traced back to the actual data source, if desired. Run analysis scripts in a robust and high-performance environment regardless of the amount of data or queries. There is also no need to create a dedicated solution for each project. Different projects can be controlled and parts can be reused: An analysis for the development of sales figures, a condition monitoring portal or even a predictive maintenance system – in one platform? With YUNA it is!

As it turns out: the human factor is of immense importance in the successful deployment of data science projects. In this context, in the next article we will take a look at how to reconcile objective and reality and literally go into the basis – the database.

Are you looking for an innovative data science platform? Then discover YUNA!

To leave a comment for the author, please follow the link and comment on their blog: R-Bloggers – eoda GmbH.

R-bloggers.com offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.

Never miss an update!
Subscribe to R-bloggers to receive
e-mails with the latest R posts.
(You will not see this message again.)

Click here to close (This popup will not appear again)