# resiudual of a regression

#### tinauser

##### New Member
Hi everybody, that's my first message on this forum...I hope you'll forgive my dumbness in statistic

I performed a fit of some X-Y data: i fit a broken line with a discontinuity at the point (Xd,0). I calculated the two slopes of the line and the point Xd minimizing the square error in a orthogonal sense.

I'm looking for some statistic to express the fact that the fit was good.

I made an analysis of residual: they are 0-mean, but theyr distribution is not normal (I performed several test). The problem is that my residual are much more clustered near the zero, so I would need a distribution similar to the gaussian but with an higher kurtosis.

the distribution of my errors can be modelized with a Student distribution with the following parameters:
nu=3 - degrees of freedom (nu = 1, 2, 3, ...)
chi=0- location parameter
eta=0.1 - scale parameter ( eta > 0 )

Do you know what kind of test I can perform?

If it's possible, I can post an example of data fit and residual histogram