The need for online handwriting recognition is rising due to the accessibility and affordability of devices such as PDAs, cellphones, and Take Notes, which can be used quickly and effortlessly. People can submit information using those devices in this recognition approach just as readily as they are accustomed to doing with a pen and paper. Using those devices has the benefit of directly storing the input data as timely, ordered stroke sequences. A significant database is essential for training and testing any handwritten recognition system and gathering information from people everywhere. Still, acquiring data from multiple sources has never been straightforward. In this paper, we are introducing a mobile application which is capable collecting online handwriting data from any user anywhere. Along with handwriting image it can store the x, y coordinate value and pen-up/pen-down value by giving the user a feelings like pen paper based method. The proposed system examined on 8800 Bangla digit including 4100 Bangla digit collected using mobile application. The system has gone though the supervised classifier like Support Vector Machine (SVM) and Random forest, Multi-Layer Perceptron (MLP) and unsupervised clustering like K-Mean using mobile application collected data as well as available Bangla digit dataset. It achieved highest accuracy 94.73% using Random forest classification. Because of its numerous applications in administrative automation, banking automation, postal automation, and human-computer interfaces, handwritten online character recognition has been a steaming research topic. This mobile application can collect the data for any language. We have focused on online Bangla Handwritten data collection as it is second most widely used language and seventh most-spoken native language throughout the world.