User Portlet
| Discussions |
|---|
| You do get a peak of the density of points at the origin. The density of points for a bivariate normal distribution with mean vector (0,0) does have a peak at (0,0). Hans introduced the distribution of the "distance" from the origin which doesn't... |
| I won't confess to a "whim" but anything from 31 and on seems to provide a smooth enough prediction (a prediction much smoother than the observed data). Also using an odd number gets one the appropriate associated time (the middle of the time... |
| It would be nice if the following worked: dist = TransformedDistribution[x1 + x2, {x1 \[Distributed] NormalDistribution[\[Mu], \[Sigma]], x2 \[Distributed] NakagamiDistribution[m, \[CapitalOmega]]}, Assumptions -> {m >= 1/2,... |
| The documentation for `FindMaximum` states in the first line: "searches for a local maximum in f, starting from an automatically selected point." So you can't necessarily get there from here. However, `Maximize` guarantees a global maximum for... |
| Same thing as my previous answer. You are attempting to fit an overparameterized model. You've included the intercept and two values for the nominal variable when there are only 2 of those parameters that are estimable. This issue should be... |
| No. Read the online reference of Discrete Multivariate Distributions under Scope. You still need dx. |
| I think the main issue is that both models are overparameterized for the available data (and secondarily better starting values are needed). A plot of the data shows just slight curvature over values of $t$ for both datasets suggesting that maybe... |
| Karl Pearson had a lot of methods. Do you mean what is described in > Pearson, K. (1901), On Lines and Planes of Closest Fit to Systems of > Points in Space |
| Cross-posted at [Mathematica StackExchange][1]. [1]: https://mathematica.stackexchange.com/questions/215688/improving-the-fitting-of-model-by-nminimize |
| Thanks! That makes much more sense. (And I do realize that despite the serial correlation, the parameter estimates from `NonlinearModelFit` can be not too bad; just their standard errors are more likely to be underestimated when ignoring the serial... |