Recently I encountered a problem with the machine traversal on a farm. It so happens that the GPS is turned ON even when the machine is traveling on the road. As shown in the figure below (just an example).
I want to figure out how do we just separate the points in the field and create a boundary for that and eliminate the points which are outside. I tried solving this using DBSCAN clustering technique.(code snippet below):
df = pd.read_csv('gps.csv')
coords = df.as_matrix(columns=['lat', 'lon'])
db = DBSCAN(eps=eps, min_samples=ms, algorithm='ball_tree', metric='haversine').fit(np.radians(coords))
It creates a cluster, but I want a solution without any clustering techniques. Maybe by creating the boundaries and checking if the boundary gets much bigger with just a few points. Anyway of trying this out in OpenStreetMap, finding whether the coordinates lie in farmland or roads? Please suggest
