Dysgraphia is characterized as a learning disability in which a person lacks writing skills that are expected for his or her age and cognitive level. Despite this, phrases such as handwriting difficulties or delayed writing performance in children are used to indicate dysgraphia. It has an effect on both the handwriting product (legibility of the written trace) and the writing process (movement that generates the trace). The dysgraphia dataset is an image-shaped dataset that contains images of children's writings that may be dysgraphia-related. These datasets were gathered through data collection techniques carried out by primary school students and children undertaking interventions at the Malaysian Dyslexia Association (PDM). Two PDM authorities then evaluated the dataset to determine whether the handwriting had potential dysgraphia or low potential dysgraphia. This dataset is used to determine the risk of dysgraphia in children. Machine learning techniques can be used to implement this data. This dataset has two types of dysgraphia levels: potential dysgraphia and low potential dysgraphia. Through the geometric characteristics and features of the writing in the image, this dataset could potentially be utilized to determine the presence of dysgraphia symptoms.