# curve fit for large data set

Need to find curve/trendline and equation that fits the large data set. Looks as in the scatter plot given below. any suggestion in a particular direction are appreciated. Any libraries in java or python or any online tools. I have tried excels trendline option, but the equation underfits my data.

• Try localized regression. If you're happy working with Python, you can use statsmodels.sourceforge.net/devel/generated/… – Jon Oct 12 '16 at 21:12
• is there any way that u know of, to get the resulting equation for the fit? – asaini Oct 12 '16 at 21:46
• Nah, I don't like to use Python for stats stuff. A simple Google search should help you out though. – Jon Oct 12 '16 at 21:59
• Notice that the spread is related to the mean, If there are no exact zeros, it's probably worth taking log(y), and looking at that. It may also be worth considering log(x), but it depends on what it actually measures as to whether that will also make sense (if the origin is arbitrary, it probably doesn't). What do the y and x variables consist of? – Glen_b -Reinstate Monica Oct 13 '16 at 0:26
• x is price , y is time – asaini Oct 13 '16 at 5:24

That looks like data you want to transform first and then maybe try a linear model. My first guess would be to try regressing $y$ against $1/x$