which algorithm will require to do this?

[abc123]

i am scientist and i have data filled with "noise". data is such for x=1 there are set of values of y e.g. for x=1 y belongs to {1,4,6,7,9,18,16,19} and i have 100 such sets each for there respected x values e.g. for x=2 y belongs to{1,5,7,4} for x=3 y belongs to {2,6,4,8,2}....for x=100 y belongs to {2,7,89,4,5} now i want to choose single value from each set of y so that this 100 values together will fit in sinusoidal curve, unfortunately whose parameters are unknown. in other language i have data for x values from 1 to 100, for each x value only one 'y' value is real rest are noise and all this 100 real values of y are in sinusoidal pattern , is any optimization algorithm exist for this?

[Mr Skeptic]

Your data is your data. Sometimes you get noisy data and sometimes not as noisy. You can't just discard data you don't like.

[insane_alien]

unfortunately there is no way of determining the right value. you'll just have to do a curve fitting