Estimation of lactation curve characteristics is essential for effective dairy sheep management, influencing nutrition, health and genetic improvement strategies. Traditional statistical methods often lack the capability to capture the complexity of milk production patterns, requiring innovative data analysis techniques. The aim of the current study is to evaluate the effectiveness of different Machine Learning (ML) algorithms - SMOreg, Linear Regression, M5 and Random Forest - in estimating key lactation curve parameters: Total Milk Yield (TMY), Peak Yield (PY) and Time to Peak Yield (TPY) under three different milk recording schemes. A total of 2,280 weekly records were used from a commercial sheep flock in Querétaro, Mexico. The results showed that the ML algorithms provided accurate estimates of lactation curve characteristics. We also found that the estimations of lactation curve characteristics were similar between the milk recording scheme using 20 records weekly (ST) and five records monthly (TF). In conclusion, the ML algorithms used can improve lactation curve estimation in dairy sheep and provide a cost-effective alternative for sheep production management by maximizing data utility while minimizing operational costs. Future research should focus on refining these algorithms for different sheep breeds and environmental conditions to further integrate them into practical farm management systems.