Possible to filter noise from measurements
Econometrician Karim Moussa introduces a new method for filtering noise from measurements in his doctoral research, with classical background noise in a phone call being just one of its many applications. Nowadays, measurements are kept for various aspects, ranging from daily step counts and heart rate to the latest stock prices, but often these measurements exhibit inaccuracies, also known as 'noise', making it difficult to draw conclusions.
Moussa's research shows that by combining two well-known statistical techniques, namely 'simulation' and 'regression', it is possible to filter noise from measurements quickly and accurately. He conducted numerous simulations to analyze the relationship between measurements with and without noise. He then used this relationship to estimate the corresponding measurement without noise based on actual measurements. Moussa emphasizes the crucial role of noise filtering in various practical applications, such as turbulence detection in airplanes and automatic vacuum cleaners, where statistical techniques involving sensors are used.
More information on the thesis