A Tool for Predicting Lawsuit Filings in the Brazilian Judiciary.

The Braziliancourt system faces significant challenges in managing the fluctuating demand for litigations, which impacts resource allocation, operational efficiency, andpolicy planning. This study applies time-series forecasting techniques to predict law suit demands, aiming to provide action able insights that can enhance the efficiency of the justice system. Five forecasting algorithms, ARIMA, Prophet, NeuralProphet, DeepAR, and N-Beats, were evaluated using historical data from two state courts: the Rio Grande do Norte State Court (TJRN) andthe Minas Gerais StateCourt (TJMG). The models wereassessedbasedonthe Root Mean Square Error (RMSE), with statistical analyses, includingthe Friedman and Nemenyitests, used to compare performance. The results showed that the Prophet and Neural Prophet algorithms out performed others, with RMSE values as low as 3.44 and 4.54 for TJMG, respectively. These findings suggest that advanced time series models can effectively forecast demand, enabling the justice system to anticipate periods of high activity, optimize resource distribution, and potentially reduce costs. Future research could extend this approach by integrating additional socio economic variables and applying these models across different segments of the justice system, opening up new avenues for exploration and development in the field. This study highlights the potential of datadriven strategies to enhance operational planning and decision-making with in the judiciary.