WHAT THE EVIDENCE SAYS
Using measured water chemistry and microbial indicators, researchers compared methods for predicting a calculated water-quality index. A random forest reported R² of 0.97 versus 0.64 for multiple linear regression; dissolved oxygen, biological oxygen demand and coliform variables remained important.
WHY IT MATTERS / OUR READ
The practical question is whether quicker data analysis improves sampling and intervention decisions. Water safety requires direct contaminant measurement and proper validation, not a model score.
CAVEATS / UNCERTAINTY
The response index is itself calculated from input variables, so fitted accuracy is not independent potability validation and might not transfer to other watersheds.
SOURCE
Discover Water · Beas/Satluj water-index model comparison
View original evidence ↗