Unlocking Logistic Regression: Maximum Likelihood Explained

Demystifying the optimization process behind fitting logistic regression models.

Slide 1: Beyond Blind Trust: Fitting Logistic Regression

Moving beyond assumptions to understand the optimization process.

    Slide 2: The Challenge: Obesity & Weight

    Applying logistic regression to a real-world example.

      Slide 3: Maximum Likelihood: The Solution

      A different approach to finding the best fit.

        Slide 4: Likelihood Calculation: Obese Mice

        Determining the likelihood for the obese group.

          Slide 5: Likelihood Calculation: Non-Obese Mice

          Determining the likelihood for the non-obese group.

            Slide 6: Log-Likelihood: Quantifying the Fit

            Turning likelihood into a single, usable metric.

              Slide 7: Finding the Optimal Fit

              The iterative process leads to the best possible line.

                Slide 8: R-squared for Logistic Regression

                The R-squared value is calculated to find the correlation between the parameters of the line.

                  Slide 9: P-value for Logistic Regression

                  The P-value for the logistic regression allows finding the best values to work.

                    Slide 10: Fitting a Line

                    The line fitting is performed via probability calculation and finding the values for the data to work.

                      Slide 11: Mouse obesity and likelihood

                      The likelihood of the obesity and not obesity of the mouses contributes to finding the line fitting.

                        Slide 12: Final Touches for a Good Line

                        Ensuring there are good values for the line and it will accurately predict new data.

                          Slide 13: The Algorithm

                          The algorithm rotates the line and checks which values is the best to fit the data for accurate prediction.

                            Slide 14: Logistic Regression's Power

                            Logistic Regression allows finding the best position for the line to predict accurate values.

                              Slide 15: Thank You

                              For your attention!