File Exchange > DataAnalysis >    Logistic Regression

Author:
OriginLab Technical Support
Date Added:
10/25/2016
Downloads (90 Days):
209
Last Update:
6/8/2017
Total Ratings:
0
File Size:
170 KB
Average Rating:
File Name:
Logistic R...on.opx
File Version:
1.1
Minimum Versions:
2017
License:
Free
Summary:

Perform binary, multinomial, and ordinal logistic regression.

Screen Shot and Video:
Description:

  • Purpose
    This tool can be used to study the relationship between a categorical response variable and independent variables by fitting with a logistic function.
    The response categorical variable can include two levels (Binary), more than two levels (Multinomial) and more than two levels with ordering (Ordinal).
    Independent variables can be continuous or categorical.
    The model can be customized.
     
  • Installation
    Download the file Logistic Regression.opx, and then drag-and-drop onto the Origin workspace. An icon will appear in the Apps gallery window.
    NOTE: This tool requires OriginPro.
     
  • Operation
    With the data worksheet active, click the app icon. A toolbar with 3 buttons will appear.Click on the appropriate button  according to the data type of the dependent variable.
     
    • Binary Logistic Regression
      1. Click the first button from the toolbar to bring up the binary_logistic dialog.
      2. Under the Input tab, set Dependent Variable and Independent Variables by using the columns in the worksheet. For Dependent Variable and Categorical Independent Variable, you can specify Reference Event and Reference Factor Level respectively.
      3. Then go to Settings tab to set the model and iterations related options. You can use Main Effects model, or customize model by yourself, including whether to use an intercept or not.
      4. Under Quantities tab, check the items you want to output, such as Fit Parameters (Odds Ratio, and Wald Test, etc.), Fit Statistics (-2 Log Likelihood, AIC, BIC, Cox Snell, McFadden's, McFadden's Adjustment, and Nagelkerke, Likelihood Ratio Test, Goodness of Fit Test, and Hosmer and Lemeshow Test, etc.), Predicted Values (Predicted Membership, Predicted Probabilities), Classification Table, Covariance Matrix, and Correlation Matrix, etc.
      5. In Diagnostic Analysis tab, you can check the corresponding checkbox to select output, including 5 residuals analysis (Regular, Standardized, Studentized, Logit, and Deviance) and Influential Cases Analysis (Cook's Distance, Leverages, and DfBeta(s)).
      6. The Plots tab is used for specifying what plots to create. Plots can be Histogram of Residual Plot, Normal Probability Plot of Residual, and Residual vs. the Order of the Data Plot. For these plots, three types of residuals is available, including Regular, Standardized, and Studentized.
      7. Finally, you can specify where to output the report and data in Output tab, and then click the OK button to generate the results.
         
    • Multinomial Logistic Regression
      1. Click the second button from the toolbar to bring up the multinomial_logistic dialog.
      2. Under the Input tab, set Dependent Variable and Independent Variables by using the columns in the worksheet. For Dependent Variable and Categorical Independent Variable, you can specify Reference Event and Reference Factor Level respectively.
      3. Then go to Settings tab to set the model and iterations related options. You can use Main Effects model, or customize model by yourself, including the decision to use intercept or not.
      4. Under Quantities tab, check the items you want to output, such as Fit Parameters (Odds Ratio, and Wald Test, etc.), Fit Statistics (-2 Log Likelihood, AIC, BIC, Cox Snell, McFadden's, McFadden's Adjustment, and Nagelkerke, Likelihood Ratio Test, etc.), Classification Table, Covariance Matrix, and Correlation Matrix, etc.
      5. Go forward to the Predicted Values tab, check the corresponding checkbox to output the predicted values. Here four options are available, and they are Estimated Response Probabilities, Predicted Membership, Predicted Probabilities, and Actual Category Probabilities.
      6. Finally, you can specify where to output the report and data in Output tab, and then click the OK button to generate the results.
         
    • Ordinal Logistic Regression
      1. Click the third button from the toolbar to bring up the ordinal_logistic dialog.
      2. Under the Input tab, set Dependent Variable and Independent Variables by using the columns in the worksheet. For Dependent Variable, you can specify the Order of Response by separating them using "|", and for Categorical Independent Variable, you can specify Reference Factor Level.
      3. Then go to Settings tab to set the model and iterations related options. You can use Main Effects model, or customize model by yourself. Here you can also choose what link function to use: Logit, Probit, Complementary Log-Log, Negative Log-log, or Cauchi.
      4. Under Quantities tab, check the items you want to output, such as Fit Parameters (such as Wald Test, etc.), Fit Statistics (-2 Log Likelihood, AIC, BIC, Cox Snell, McFadden's, McFadden's Adjustment, and Nagelkerke, Likelihood Ratio Test, Equal Slopes Test, etc.), Covariance Matrix, and Correlation Matrix, etc.
      5. Go forward to the Predicted Values tab, check the corresponding checkbox to output the predicted values. Four options are available: Estimated Response Probabilities, Predicted Membership, Predicted Probabilities, and Actual Category Probabilities.
      6. Finally, you can specify where to output the report and data in Output tab, and then click the OK button to generate the results.

Updates:

v1.1: fixed bug for no continuous independent specified

Reviews and Comments:

Be the first to review this File Exchange submission.