In the last twenty years, I’ve seen an explosion in predictive analytics. From predicting customer churn in marketing to forecasting equipment failures in manufacturing, predictive analytics brings the power of AI and data to divining the future.
Firstly, let’s define the term. Predictive analytics is usually compared to descriptive analytics. Predictive analytics is a subfield of data science that focuses on using statistical and machine learning techniques to make predictions about future outcomes. Descriptive analytics, on the other hand, focuses on analyzing and interpreting historical or real-time data.
Take an example from manufacturing. A data scientist could use descriptive analysis to look at data on production rates, defects, and equipment uptime to understand how different factors impact the efficiency of the manufacturing process. The data scientist would then use this analysis to make recommendations for improving the process, such as by identifying bottlenecks or inefficiencies that could be addressed.
Conversely, a data scientist could use predictive analysis to build a model that predicts the likelihood of an equipment breakdown based on factors such as the equipment's age, usage history, and maintenance records. The model would be used to predict when the equipment is likely to need maintenance or repair, allowing the manufacturer to schedule these activities in advance and avoid unexpected downtime.
When companies start their analytic journey, most of their focus is on descriptive analytics— answering the question “What happened?”. But as companies become more analytically mature, the focus shifts to predictive analytics, answering the question, “What’s going to happen?
In the examples above, descriptive analysis allows for statements like, “On average, 26 pieces of equipment suffer breakdowns per week at plant XYZ.” While predictive analysis allows for statements like “Machine 153 has a high chance of breakdown in the next four hours and needs to be examined at the soonest scheduled maintenance period,” The difference between these two statements is the specificity of the action. While it’s possible to get some value from knowing at the end of the month if the breakdowns were better or worse than average, the real value comes from the second statement, the ability to do something about it!
When companies start their analytic journey, most of their focus is on descriptive analytics—answering the question “What happened?”. But as companies become more analytically mature, the focus shifts to predictive analytics, answering the question “What’s going to happen?”.
This is primarily because predictive analytics is more difficult than descriptive analytics. It involves more complex mathematical and statistical models and more sophisticated data analysis techniques. Additionally, the process of developing a predictive model can be more time- and resource-intensive than the process of summarizing and visualizing historical data. To do this, it often requires technical expertise in advanced methods in data science and statistics, as well as large data sets, powerful computing resources, and sophisticated software tools.
Predictive analytics is an important tool in many industries because it allows organizations to make data-driven decisions by analyzing historical data and identifying patterns. With the ability to predict future outcomes and behaviors, organizations can take proactive measures to improve their operations, reduce costs, and increase efficiency. At Ford, we’ve used predictive analytics in areas as diverse as manufacturing (in fact, my team is working on the equipment example mentioned earlier), as well as use-cases around optimization of manufacturing assets, predicting supply chain disruptions, and improving shipping efficiency.
Overall, predictive analytics is a powerful tool that can help organizations make better decisions, improve their performance, and gain a competitive advantage.
