For a class called Quantitative Engineering Analysis 3, my small group of three created a phone typing speed estimator algorithm. It takes accelerometer data from a phone collected during a typing test, performs a Fourier transform on the data to identify its most prevalent frequencies, compares it to a calibration curve we previously generated using test data and outputs an estimated typing speed in words per minute.
Testing setup: one phone in front which is being typed on for the test, another held behind to collect accelerometer data.
Plot of test data points used to develop the trend line of the most prevalent frequency found through Fourier transform analysis vs the WPM of that trial. The calculated trend line for this data set is also included.
The results of our process using a test dataset from each of the group members, along with the percent error in the output of the algorithm.