Logistic Regression, Basic
This function builds a binary logistic regression model using a formula builder. BioStat Prime uses glm function passing the parameter family =binomial(link='logit'). BioStat Prime displays a summary of the model, analysis of variance tables and McFadden R2. User can score the model by selecting the model created on the top right hand corner of the main application screen and select the Score button. User can choose to display a confusion matrix and a ROC curve
To analyse it in BioStat Prime user must follow the steps as given.
- Steps
Load the dataset -> Click on the Model Fitting tab in main menu -> Select Regression -> This leads to analysis techniques, choose Logistics, Basic -> There will appear a dialog -> Select the model name, dependent variables and populate the formula builder in the dialog -> Check the radio buttons to display a plot in the output -> Finally execute the plot and visualise the output in output window.
Arguments
- depVar
Name of the dependent variable. If we have a dataset cars, with a variable class that we want to predict (dependent variable is class) enter class
- indepVars
Names of the independent variable, separated by +. If we have a dataset cars, with independent variable horsepower, enginesize, specify horsepower+enginesize). Categorical variables are automatically dummy coded.
- data
Name of the dataframe. When you open data frames or datasets e.g. csv, Excel files, SAS files in BioStat Prime, they are named Dataset1, Dataset2, Dataset3 So enter data=Dataset1.
This can cause a mismatch as NA values are NOT automatically removed from the weighting variable. In this situation you will see the error variable lengths differ (found for (weights))