Data has shapes, and shapes may provide insights to data modeling and information extraction. Topological data analysis (TDA) paves new avenues in the evaluation of biomedical data, where algebraic-topological tools are used for knowledge discovery. In the present work, we apply TDA for personalized electrocardiographic signal classification toward arrhythmia analysis. First, to facilitate the TDA, signal samples are converted into point clouds using phase space reconstruction. Topological techniques are then used to extract the persistence landscapes from the point clouds as features used in the subsequent arrhythmia classification. The proposed persistence landscape based feature learning method is robust to the size of the training set. With only 20% of the full training dataset, it achieves a 100% accuracy for normal heartbeats, 97.13% for ventricular beats, 94.27% for supra-ventricular beats, and 94.27% for fusion beats. Thus the method can be trained for a single individual, allowing for personalized analysis systems. With the present study, we show that TDA could be a useful tool for biomedical signal analysis, with potential application in the personalized data processing.