The paper investigates the ability of well-established Statistical Process Control Techniques to efficiently monitor the performance of a Machine Learning classification algorithm and signal a possible deterioration of such performance. A timely and justified reaction to such signals may decrease the overall cost of maintaining the algorithms. The paper highlights the weaknesses of the current popular techniques and proposes a low-cost methodology to monitor the ongoing performance of classification algorithms.

Recommended Citation: Samokhvalov I., Kline D., Guinn C., Vanajakumari M., (2022). Detecting Shifts in the Performance of the Sentiment Classification Algorithm Using Twitter Feed Related to a Public Company. UNCW MS CSIS Proceedings. V. 16 , N. 0 .