Here we present a novel machine-learning approach to predict protein aggregation propensity (PAP) based on logistic regression (LR). We used a dataset of hexapeptides with known aggregation tendencies and eight physiochemical features to train and test the LR model. Also, we evaluated the performance of the LR model using F-measure and Matthews correlation coefficient (MCC) as metrics and compares it with other existing methods. Moreover, we investigated the effect of combining sequence and feature information in the prediction. In addition, the overall performance of the concluded method was higher than the other known servers, for instance, Aggrescan, Metamyl, Foldamyloid, Pasta 2.0.