Big Data: checklist for the implementation of analytical projects

Big Data analytics projects are high on the list of IT priorities for many organizations seeking to maximize the business benefits of all the data (structured, unstructured, and semi-structured) that enters their systems.

But as with any initiative that can pay big , the risks are also high. This is the case of the implementation of Big Data , which makes the planning and management of deployments all the more essential.

There are good methods, but also bad ones. The following is a list of steps that project leaders need to take to get their Big Data programs right: the one that will enable the company to achieve the expected business value and a good return on investment. .

Look for business references that offer a serious business plan.

With all the hype surrounding Big Data, it is likely that executives wanting to sponsor a project will be numerous.

The main criteria you must use to select them is their ability to develop a set of clear business goals with realistic deadlines. By having a clear idea of ​​the results your company is aiming to achieve, you will be able to determine the scope of the data management and analytics systems you will need to build alongside the core technology to install.

If a project starts without this type of framing, there is a good chance that it will escape control by trying to do too much and too quickly.

Integrate learning, but also mistakes, into the project plan.

Big Data analytics will introduce new technologies, techniques and methodologies into your organization ( such as NoSQL or Hadoop ), and will likely require new skills or evolve the ones businesses already have .

In addition, Big Data technologies are constantly evolving and a huge amount of custom development work is often required. The new skills required are still very rare , whether they are IT developers and data scientists or other analytics professionals who will lead the data analysis work.

As a result, your project team will learn on the job, and business executives and users will gradually discover what big data analytics really means to them. Set the project schedules and budgets with a long learning curve, not to mention the inevitable mistakes that will mark that learning.

Use agile application development .

Since everyone will likely need substantial training and the detailed business needs may change along the way, Agile development methodologies are better suited to Big Data analytics applications than standard cascading approaches.

An Agile approach provides iterative small fragment functionality and facilitates rapid changes in development plans. It is better adapted when many uncertainties remain.

This approach must be accompanied by a visible and transparent process of change management , as well as regular communication with project leaders and participants about progress and the inevitable changes.

Plan the least activity .

One of the proven project management rules, especially in software development, is that the job will fill any available time slot.

As a leader in a Big Data analytics project, you will certainly have the chance to have an extremely enthusiastic community of executives and employees looking for information to guide operational strategies and tactics. . While learning on the job and openness to change is part of the process, you need to build on this enthusiasm by organizing activities in tight time slots, so that the Big Data initiative can move forward. participants do not become discouraged if they are blocked on certain tasks.

Treat data scientists as artists.

The data-scientists and other analysts have a key role to extract business knowledge from Big Data. Generating this knowledge, using applications such as predictive analytics and data mining, is an incremental and iterative process.

The data-scientist develops an analytical model, tests it, refines it and validates it, finally applying it and publishing the results internally.

In doing so, it can test tens or hundreds of variables using various statistical methods. The term data-scientist is somewhat misleading: creating analytical knowledge is as much about science as it is about art. Treat data scientists as talented artists rather than ordinary workers if you want to boost their productivity and achieve better results.

Set realistic expectations and proactively manage them .

In companies that are new to Big Data, technology providers can set ambitious goals by ensuring that Big Data tools are easy to use and by citing other companies that have leveraged them. significantly increase their business value.

It is important to keep in mind that most of the pioneers of Big Data systems were large Internet companies with strong expertise that have very often played a prominent role in the development of Hadoop and other Big Data technologies.

If you let your goals slip out of control and can not reach them afterwards, implementing your big data analysis may be perceived as a failure, no matter how much value it actually produces. for the company. Set realistic goals from the start, and continue throughout the project.

Conclusion

Of course, launching a big data analytics project involves considerable risk while making big money. But by adopting rigorous project management practices as much as possible, project managers and their teams will limit the negative aspects and turn deployments into real business opportunities.

Leave a Reply

Your email address will not be published. Required fields are marked *