Least squares curve fitting


    n. A way of fitting a polynomial to observed data. Suppose you want to build a device to measure the amount of protein in a sample of grain. There’s a complicated relationship between protein content and the amount of IR light reflected from the grain at dozens of frequencies. Unfortunately, the relationship isn’t well understood, so no simple algorithm exists to predict protein content. Instead, instrumentation vendors calibrate each machine by reading hundreds of grain samples with known protein levels, then doing a least squares curve fit to determine the coefficients of a polynomial that matches the observed IR radiation to known protein contents. The resulting polynomial can then predict the protein content of unknown samples.

    Previous articleLeast significant byte
    Next articleLeast squares error


    Please enter your comment!
    Please enter your name here