In this course small groups of students worked to improve the existing data analysis developed by previous classes. The goal of the pipeline is to classify radio-signals from the Greenbank Radio Telescope, the world’s largest radio telescope, as technosignatures of extraterrestrial civilizations.
We collected terabytes of data, and using a pipeline made by students from past years this class has been offered, reduced the number of candidate signals to just over a million candidate signals. Some criteria we used to reduce the number of signals was considering the drift rate of the signal, frequency of the signal (in ), and if we observed a particular signal in other observations. Many GPS satellite operate at known frequencies, and orbit the Earth at known rates and heights. Doing some math we can determine whether the signals we observed in some observations with the telescope were from man made objects.
My group wrote a Python script that would plot various features of a signal against one another. We used libraries like numpy, matplotlib, and scikit learn. The various signal features we analyze were frequency, drift rate, and signal to noise ratio. Furthermore, we also clustered our data using K-Means to see if we could observe any similarities within clustered signals.