You cannot select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
78 lines
2.8 KiB
Plaintext
78 lines
2.8 KiB
Plaintext
7 months ago
|
.-
|
||
|
|
help for ^discrim^ (STB-x)
|
||
|
|
.-
|
||
|
|
|
||
|
|
Dichotomous Discriminant Analysis
|
||
|
|
---------------------------------
|
||
|
|
|
||
|
|
^discrim^ grpvar [varlist] [^, p^redict^ a^nova^ g^raph^ d^etail^ k^eep]
|
||
|
|
|
||
|
|
Description
|
||
|
|
-----------
|
||
|
|
^discrim^ allows the user to perform a discriminant analysis on a Bernoulli
|
||
|
|
distributed response of grouping variable (1/0).
|
||
|
|
|
||
|
|
Output includes table with the following statistics and their global
|
||
|
|
designation for post-estimation use when applicable:
|
||
|
|
|
||
|
|
1. Observations [S_E_obs] 2. Independent vars [S_E_var]
|
||
|
|
3. Obs: group==0 [S_E_ob0] 4. Obs: group==1 [S_E_ob1]
|
||
|
|
5. Centroid 0 [S_E_cn0] 6. Centroid 1 [S_E_cn1]
|
||
|
|
7. Grand centroid [S_E_cng] 8. R-square [S_E_r2]
|
||
|
|
9. Mahalanobis [S_E_mah] 10. Eigenvalue [S_E_eig]
|
||
|
|
11. Canon. Correl [S_E_cc] 12. Eta squared [S_E_e2]
|
||
|
|
13. Wilk's lambda [S_E_lam] 14. Chi-square [S_E_chi]
|
||
|
|
|
||
|
|
Output also includes a table displaying both the Discriminant Function and
|
||
|
|
Unstandardized Coefficients.
|
||
|
|
|
||
|
|
Cases in which a variable has a missing value are dropped from estimation.
|
||
|
|
The ^keep^ command does not restore such cases to active memory.
|
||
|
|
|
||
|
|
Options
|
||
|
|
-------
|
||
|
|
^predict^ provides a confusion matrix of actual vs predicted group cell
|
||
|
|
counts. Also included is a listing of the following percentages:
|
||
|
|
1. Correctly predicted 2. Model sensitivity
|
||
|
|
3. Model specificity 4. False Positive
|
||
|
|
5. False Negative
|
||
|
|
|
||
|
|
^anova^ provides an ANOVA of discriminant scores vs the group variable.
|
||
|
|
Bartlett's test for equal variances in included.
|
||
|
|
|
||
|
|
^graph^ provides a classification graph showing correctly and incorrectly
|
||
|
|
classified cases.
|
||
|
|
|
||
|
|
^detail^ creates and lists the following for each retained case:
|
||
|
|
1. Group : actual group value
|
||
|
|
2. PRED : predicted group value
|
||
|
|
3. DIFF : a star indicating misclassified cases
|
||
|
|
4. LnProb1 : logistic probability of grouyp 1 membership
|
||
|
|
5. DscIndex : discriminant index
|
||
|
|
6. DscScore : discriminant score
|
||
|
|
|
||
|
|
^keep^ allows the user to keep the variables created in the Detail option.
|
||
|
|
The data set in memory is altered, so care must be taken when saving the
|
||
|
|
resultant file to disk.
|
||
|
|
|
||
|
|
Examples
|
||
|
|
--------
|
||
|
|
. ^discrim foreign price mpg length^
|
||
|
|
|
||
|
|
. ^discrim foreign price mpg length, p a gr d^
|
||
|
|
|
||
|
|
. ^discrim foreign price mpg length, detail keep^
|
||
|
|
|
||
|
|
Also see
|
||
|
|
--------
|
||
|
|
STB-5 (Jan1992): smv3 - Hilbe, "Regression based dichotomous discriminant
|
||
|
|
analysis".
|
||
|
|
Selvin, Steve (1995, 1996) "Practical Biostatistical Methods", Chap 6.
|
||
|
|
Explanation and use of the 1st version of the ^discrim^ command.
|
||
|
|
|
||
|
|
Additional help
|
||
|
|
---------------
|
||
|
|
Joseph Hilbe, Dept of Sociology, Arizona State University
|
||
|
|
Email: atjmh@@asuvm.inre.asu.edu
|
||
|
|
|