This paper displays a novel technique that allows tuning and predicting better values for A3 offset handover parameter for both inter-frequency and intra-frequency. The proposed work takes effect by combining statistical trend analysis (change point detection) and supervised machine-learning technique where a planned parameter change will trigger a corresponding change in handover success rate (HOSR). That change will be captured by statistical trend analysis in order to feed the class target function of the supervised machine-learning model for the data-set created. This process is introduced to help the model classifier predict better values for the A3 parameter offset for degraded cells with a low handover success rate. Using domain knowledge, the authors leveraged 20 days of data captured from a mobile operator serving more than 2 million subscribers to use for our model analysis and development. Also supervised machine learning model is trained to predict enhancement or degradation of handover success rate with proposed values of A3 offset parameter change. In fact, The adopted methodology can be extended to tune other logical parameters in order to accelerate machine learning implementation in key performance indicators(KPI) prediction and radio frequency (RF) parameter tuning.