Based on "Using Educational Data Mining to Predict Students' Academic Performance for Applying Early Interventions" by Sarah Alturki, Nazik Alturki, and Heiner Stuckenschmidt (2021), educational data mining (EDM) can reveal many useful insights but also has limitations.

What EDM can tell us:

It helps predict students' performance early by identifying who might excel or struggle. For example, it can show if a student might need extra help with core courses like programming or databases.
It can also spot high achievers early on, so schools can offer scholarships or special programs.
Patterns in grades and the number of failed courses are strong indicators of future performance, which allows teachers to step in before problems escalate.

What EDM cannot tell us:

It only shows what is happening—like a drop in grades—but not why it is happening. Personal challenges, motivation, or external factors are harder to capture through numbers.
Some things that seemed important, like English skills or orientation programs, turned out not to be good predictors in this study, meaning data can sometimes surprise us.

In short, EDM helps teachers and schools take action sooner, but it works best when combined with personal insights into students’ lives and needs.

reference:

Alturki, S., et. al., 2021. “Using Educational Data Mining to Predict Students Academic Performance for Applying Early Interventions. doi:10.28945/4835
https://www.researchgate.net/publication/353419538_Using_Educational_Data_Mining_to_Predict_Students'_Academic_Performance_for_Applying_Early_Interventions