# Model accuracy without over fitting data

I was going through machine learning topics and came across term "Over-fitting of data", So question i want to ask, is it possible for any model to have 100% accuracy without over-fitting the data ?

• Yes, it is possible. – Carl Nov 22 '16 at 6:47

Yes, it is not only possible but happens when 1) the number of parameters equals the sample size, and 2) a real solution is possible otherwise $n$-tuple complex solutions are the result (and assuming that we know how to find those complex solutions, which is not for the faint hearted). Other scenarios of exact solutions occur when the data is exact, i.e, noiseless, and we are regressing with the exact same model as our data obeys.