1 | ## Regression Models
|
2 |
|
3 | ## Instance Functionality
|
4 |
|
5 | ### ols( endog, exog )
|
6 |
|
7 | What's the `endog`, `exog`?
|
8 |
|
9 | Please see:
|
10 |
|
11 | http://statsmodels.sourceforge.net/stable/endog_exog.html
|
12 |
|
13 | `ols` use ordinary least square(OLS) method to estimate linear model and return
|
14 | a `model`object.
|
15 |
|
16 | `model` object attribute is vrey like to `statsmodels` result object attribute
|
17 | (nobs,coef,...).
|
18 |
|
19 | The following example is compared by `statsmodels`. They take same result
|
20 | exactly.
|
21 |
|
22 | var A=[[1,2,3],
|
23 | [1,1,0],
|
24 | [1,-2,3],
|
25 | [1,3,4],
|
26 | [1,-10,2],
|
27 | [1,4,4],
|
28 | [1,10,2],
|
29 | [1,3,2],
|
30 | [1,4,-1]];
|
31 | var b=[1,-2,3,4,-5,6,7,-8,9];
|
32 | var model=jStat.models.ols(b,A);
|
33 |
|
34 | // coefficient estimated
|
35 | model.coef // -> [0.662197222856431, 0.5855663255775336, 0.013512111085743017]
|
36 |
|
37 | // R2
|
38 | model.R2 // -> 0.309
|
39 |
|
40 | // t test P-value
|
41 | model.t.p // -> [0.8377444317889267, 0.15296736158442314, 0.9909627983826583]
|
42 |
|
43 | // f test P-value
|
44 | model.f.pvalue // -> 0.3306363671859872
|
45 |
|
46 | The adjusted R^2 provided by jStat is the formula variously called the 'Wherry Formula',
|
47 | 'Ezekiel Formula', 'Wherry/McNemar Formula', or the 'Cohen/Cohen Formula', and is the same
|
48 | as the adjusted R^2 value provided by R's `summary.lm` method on a linear model.
|