Re: 2 obvious
- To: mathgroup at smc.vnet.net
- Subject: [mg115231] Re: 2 obvious
- From: Darren Glosemeyer <darreng at wolfram.com>
- Date: Tue, 4 Jan 2011 18:50:01 -0500 (EST)
On 12/23/2010 2:55 AM, Francisco Gutierrez wrote:
> Dear Group:
> Ok, here comes a really silly question. After v. 7, the statistical capacities of Mathematica have been substantially boosted. However, I haven't been able to interpret the simple command LogitModelFit. THe problem is that the documentation only offers examples with two independent variables. I have not managed to find how can n more independent variables can be inserted so that the command works. Can somebody send me a simple and clear example?
>
> Let this be a pretext to thank all the incredibly useful help the members of this list have generously provided to me and others. Happy holidays,
> Fg
>
>
>
>
Hi Francisco,
The data argument is like for other fitting functions. The independent
(predictor) variables are the first n-1 elements in each data point and
the last element in the data point is a dependent (response) variable.
Here is a list of 5 data points each having two predictors and one response.
In[1]:= data = {{10, 4, 0.26}, {8, 3, 0.04}, {2, 0, 0.17}, {4, 8, 0.09},
{9, 4, 0.83}};
This treats the predictors as x and y in the fitting
In[2]:= lm = LogitModelFit[data, {x, y}, {x, y}];
and here we get the fitted function and a table of parameter information
for the fitting.
In[3]:= lm[{"BestFit", "ParameterTable"}]
1
Out[3]= {-------------------------------------,
2.384 - 0.236963 x + 0.0664776 y
1 + E
> Estimate Standard Error z\[Hyphen]Statistic
P\[Hyphen]Value}
1 -2.384 3.41536 -0.698022 0.485163
x 0.236963 0.388785 0.609496 0.542196
y -0.0664776 0.529535 -0.125539 0.900097
I hope this helps.
Darren Glosemeyer
Wolfram Research