This work presents Apis-Prime, a hybrid deep learning model for soft sensing and time series forecasting, to estimate the daily weight variations of honeybee hives. Apis-Prime improves the state-of-the-art of earlier proposed WE-Bee (Anwar et al., 2022), and also helps optimize the beehive monitoring systems for the task of daily weight variation estimation. Weight variations of a honeybee hive are the most important indicator of hive productivity, and the health and strength of a bee colony. Currently, precise measurement of the weight of a hive requires an expensive weighing scale under each hive. On the other hand, sensors deployed inside the hive are cheaper than a weighing scale, and are shielded from the extreme weather variations outside the hive. In this work, honeybee activity is monitored using data from sensors inside the hive, along with monitoring the information related to the seasons, time of the day, external weather and the size of hive. Apis-Prime’s deep learning algorithm is based on two self-attention encoders, which collectively transform the sensor data into daily weight variations of the hive. Two parallel encoders simultaneously pay attention to time based relationships and feature based relationships within the daily sensor data, and generate daily hive weight estimates with a better accuracy. Comparison shows an average error of 19.7 grams/frame for Apis-Prime, compared to 21.05 grams/frame for the earlier proposed model WE-Bee. For system optimization, this work uses the attention weights of trained encoders of Apis-Prime to evaluate the sensors features collected by the monitoring system. This evaluation is used to identify and remove the unnecessary sensors/features from the dataset, reducing the number of features from 36 to 23, hence providing a significant optimization of cost, power and data-bandwidth. We provide a performance analysis of beehive weight estimations by Apis-Prime using the complete, as well as the optimized dataset on 2,170 days of beehive sensor recordings. Equally good results of daily weight estimations using the optimized feature set demonstrate the efficacy of proposed model for optimization of beehive monitoring system for the task of hive weight estimation.