Identifying dance gesture is a challenging pattern recognition problem since it involves issues like occlusion, close similarities between different dance gestures, illumination problem, etc. Dance gesture identification involves analyzing human gesture in circumstances where creative expression is conveyed through gestures. This study includes creation of our own dataset which comprises of 2305 images of hand gestures from Sattriya Dance. Most of the existing works on Sattriya Dance hand gestures used image pixel-based and statistics-based features. From the literature review, it is seen that there is a lack of research focusing on Sattriya Dance gestures that utilize hand skeleton information. In our work, we are incorporating positions of different skeleton points of hand to classify hand gestures. Here, we have used three features. It includes distance, angle and eccentricity. Various existing classification techniques were tested on our dataset, which are showing promising results.