Businesses are looking for predictive analytics that promise to increase sales, reduce costs, prevent fraud and streamline operations.
Yet most organizations fail to achieve the desired results. A study by the MIT Sloan Management Review and the Boston Consulting Group found that only 10% of companies have reaped significant financial benefits from their AI investments. And in a investigation According to Rexer Analytics, only 22% of data scientists said their new initiatives are generally deployed and operationalized across the company.
Many predictive machine learning projects fail because they focus too much on the technology alone instead of advancing the technology as a strategic business project, according to Eric Siegelconsultant and former professor at Columbia University and the University of Virginia.
In his new book, “The AI Handbook: Mastering the Rare Art of Deploying Machine Learning“, Siegel argues that organizations fail to see the value in AI because they lack an effective business paradigm for executing machine learning projects. Since most machine learning projects are highly technical , they often fall into the domain of experienced data science professionals. The result is a disconnect between the data experts who prepare data and develop and operate AI models and the business stakeholders responsible for managing operations at scale. who can benefit from predictive information.
“Focusing so much on the science of modeling rather than deployment is like being more excited about rocket science than actually launching the rocket,” Siegel said in a recent statement. MIT Sloan Management Review Webinar. “That’s where we are today.”
To drive success, Siegel said, companies need a standardized playbook for machine learning projects that is accessible to professionals and can help them participate in the predictive analytics project lifecycle.
Otherwise, “both parties point at each other and say, ‘Executing and managing this process at the enterprise level is not my job,'” he explains. “It’s sitting in no-man’s land, and it’s the last remaining ingredient before we get success and larger-scale deployment.”
6 steps to launching machine learning projects
To bridge the gap, Siegel advocates what he calls “BizML,” a set of business practices for running predictive machine learning projects.
He outlined six steps to foster collaboration between business and technical stakeholders throughout all phases of machine learning deployment:
Establish the deployment goal. To derive real value from machine learning, businesses need a defined value proposition that details the impact of the technology on their operations. Data scientists cannot do this in isolation. It is important that business stakeholders who are fully aware of the problems and opportunities are technologically savvy enough to participate in setting realistic goals.
Establish the prediction goal. Even though modeling and preaching involve complex mathematics, business goals must be kept in mind. Business users should have a semi-technical understanding of the technology so they can share their specific domain knowledge while defining what the machine learning model is supposed to predict for each use case.
Establish the right measurements. Determine important benchmarks to follow during model training and deployment. Additionally, identify the performance levels that must be achieved for the machine learning project to be considered a success. Typically, most machine learning projects rely on technical metrics like precision, recall, or accuracy. Organizations should focus on business metrics such as profit, ROI, savings and customer acquisition, Siegel said.
Prepare the data. Define what the training data should look like and make sure it is in the desired format. This critical step is non-negotiable because it is the key to achieving high-value results, Siegel said.
Train the model. Then, the prepared data is used to train and generate a predictive model. Data scientists are leading the charge in this area, but there is always opportunity to make additional business contribution.
Related Articles
Deploy the model. Use the model to generate predictive scores and, in turn, use these scores to improve business operations. It is also important to maintain the models through continuous monitoring and periodic updating.
Although the last three steps are more technical than the first three, they all require extensive collaboration between technology and business stakeholders. Building bridges to connect the two camps requires investment and commitment to change management best practices to ensure adequate understanding of machine learning by business stakeholders.
“Change management challenges are generally not new, but when it comes to machine learning projects, the need to manage change wisely is often overlooked,” Siegel said. “Machine learning delivers a rocket, but managers still need to command its launch. »
