Falls are one of the main causes of injury and injury-related deaths in the older demographic sector. During the last decade, several solutions have been proposed for fall detection or prevention, but also to assist old adults in their mobility. The main concern is to provide means to improve their quality of life, allowing them to live in community as long as possible, but also to help maintaining public health systems as sustainable as possible. In this paper, we compare two different methodologies that aim to define walking patterns in 42 adults aged 22 to 94 years that participated in a pilot study. Two methodologies are compared: one focused on the whole exercise performance, while the other tries to identify strides and categorize them. Both proposals use a technique based on the bag-of-words approach.