The weather department predicted rain yesterday, but it did not rain

Why

Kartik has a real data point — the forecast genuinely did not match what happened, and noticing that mismatch is fair. The trouble is stretching one miss into 'always useless', which is a much bigger claim than one wrong day can support. Weather forecasts are built from patterns in huge amounts of data (temperature, pressure, humidity, wind, satellite images) to give the MOST LIKELY outcome, not a guarantee — the atmosphere is complex enough that even excellent forecasting gets some days wrong, especially several days ahead. What actually tells you whether forecasting is useful is its accuracy over MANY predictions, not one single day, the same way one wrong answer doesn't prove a generally careful student is bad at the subject. Test: keep a diary comparing the forecast to the actual weather every day for a month — you'll likely find it right far more often than wrong, even though some days will miss. No, one wrong forecast does not prove that: judging a forecasting system needs to look at its accuracy over many predictions, not a single day it missed.