In this study, natural language processing was performed using a dataset of comments collected from a Turkish movie review website. The data used in the study were commented on and rated by users on the Turkish movie review website. These data were rated on a scale from 0.5 to 5.0. Feature selection metrics are widely used in the f ield of statistics. After the data was obtained, the contribution of LR to its success was determined by using the discriminative feature of feature selection metrics in comments belonging to various categories with Logistic Regression (LR) to meet the requirements of this research. In the proposed system design, an accuracy rate of 89% was achieved when classifying only positive and negative categories. Based on the findings in the literature review, the proposed system design proves that feature selection metrics can be successfully used in sentiment analysis. It is believed that this proposed new system design will bring a new perspective to the field of sentiment analysis.