You were previously given a broad overview of logistic regression. This included two separate packages for creating logistic regression models. In this lab, you'll be investigating fitting logistic regressions with statsmodels.
You will be able to:
- Implement logistic regression with statsmodels
- Interpret the statistical results associated with regression model parameters
The statsmodels example we covered had four essential parts:
- Importing the data
- Defining X and y
- Fitting the model
- Analyzing model results
The corresponding code to these four steps was:
import pandas as pd
import statsmodels.api as sm
#Step 1: Importing the data
salaries = pd.read_csv("salaries_final.csv", index_col = 0)
#Step 2: Defining X and y
x_feats = ["Race", "Sex", "Age"]
X = pd.get_dummies(salaries[x_feats], drop_first=True, dtype=float)
y = pd.get_dummies(salaries["Target"], dtype=float)
#Step 3: Fitting the model
X = sm.add_constant(X)
logit_model = sm.Logit(y.iloc[:,1], X)
result = logit_model.fit()
#Step 4: Analyzing model results
result.summary()
Most of this should be fairly familiar to you; importing data with Pandas, initializing a regression object, and calling the fit method of that object. However, step 2 warrants a slightly more in depth explanation.
Recall that we fit the salary data using Race, Sex, and Age. Since Race and Sex are categorical, we converted them to dummy variables using the get_dummies() method. The get_dummies() method will only convert object and category data types to dummy variables so it is safe to pass Age. Note that we also passed two additional arguments, drop_first=True and dtype=float. The drop_first=True argument removes the first level for each categorical variable and the dtype=float argument converts the data type of all of the dummy variables to float. The data must be float in order to obtain accurate statistical results from statsmodel. Finally, note that y itself returns a pandas DataFrame with two columns as y itself was originally a categorical variable. With that, it's time to try and define a logistic regression model on your own!
Import the data stored in the file titanic.csv.
# Your code hereFor your first foray into logistic regression, you are going to attempt to build a model that classifies whether an individual survived the Titanic shipwreck or not (yes it's a bit morbid). Follow the programming patterns described above to define X and y.
# Your code hereNow with everything in place, initialize a regression object and fit your model!
Statsmodels was unable to fit the model due to some Linear Algebra problems. Specifically, the matrix was not invertible due to not being full rank. In layman's terms, there was a lot of redundant, superfluous data. Try removing some features from the model and running it again.
# Your code hereGenerate the summary table for your model. Then, comment on the p-values associated with the various features you chose.
# Your code hereCreate a new model, this time only using those features you determined were influential based on your analysis in step 4.
# Your code hereWell done! In this lab, you practiced using statsmodels to build a logistic regression model. You then reviewed interpreting the results, building upon your previous stats knowledge, similar to linear regression. Continue on to take a look at building logistic regression models in Sci-kit learn!