Predicting customer churn is essential for telecommunications companies to maintain profitability. However, training models on historical models leads to performance degradation when they are applied to future conditions—a phenomenon known as concept drift. We employ a sliding-window approach that separates the training and testing time windows, creating a future-based “true test”. Using unique real data, we show that a CatBoost classifier model trained on older data can remain relevant when new, unseen intervals are used. A key innovation of our work is the use of 40-day “partial churn” labels; a model trained on these labels accurately predicts 90-day churn by simply adjusting the decision threshold. Out of the six modeled scenarios, in the main realistic scenario, CatBoost retained an accuracy above 0.798 and an F1 of near 0.704, reflecting its robustness even under real-world delays and potential drift. Overall, our findings emphasize that models do not necessarily “expire” with time; rather, their performance varies according to when they are tested. This research underscores the importance of a truly future-based evaluation (instead of artificial splits) and offers practical guidance for earlier churn detection when facing real-world data delays.
A detailed analysis of absorbing boundary conditions for the linear Schrodinger equation is presented in this paper. It is focused on absorbing boundary conditions that are obtained by using rational functions to approximate the exact transparent boundary conditions. Different strategies are investigated for the optimal selection of the coefficients of these rational functions, including the Pade approximation, the L2 norm approximations of the Fourier symbol, L2 minimization of a reflection coefficient, and two error minimization techniques for the chosen benchmark problems with known exact solutions. The results of computational experiments are given and a detailed comparison of the efficiency of these techniques is presented.
8 April 2011 in Mathematical Modelling and Analysis
We consider coupled nonlinear equations modelling a family of travelling wave solutions. The goal of our work is to show that the method of internal averaging along characteristics can be used for wide classes of coupled non-linear wave equations such as Korteweg-de Vries, Klein – Gordon, Hirota – Satsuma, etc. The asymptotical analysis reduces a system of coupled non-linear equations to a system of integro – differential averaged equations. The averaged system with the periodical initial conditions disintegrates into independent equations in non-resonance case. These equations describe simple weakly non-linear travelling waves in the non-resonance case. In the resonance case the integro – differential averaged systems describe interaction of waves and give a good asymptotical approximation for exact solutions.