This study analyzes the performance of the Logistic Regression algorithm in facial image classification using a combination of color and shape features. These features are used to extract key information to distinguish individuals. Logistic Regression was chosen for its simplicity and ability to produce probabilities, despite the availability of more complex algorithms. The dataset includes variations in facial expressions and lighting conditions. The results show that combining color and shape features improves classification accuracy to 80%, with a recall of 88% and a precision of 75.86%. However, the algorithm shows limitations when dealing with high-complexity data. This study contributes to the development of more efficient facial recognition systems and offers insights into the strengths and limitations of Logistic Regression in image classification tasks.