While regression algorithms within Machine Learning (ML) is claimed efficient in crime rate detection, very few studies shed light in crime rate prediction, especially in Bangladesh. Moreover, previous experiments conducted with exotic data, which enriches the ML field, however lacks in providing first hand experiment. This is, because crime rate is varied from country to country. Realizing these two important gaps in the field of crime rate prediction using ML, in this study I collected crime rate data from police.gov.bd- a Bangladeshi police website; and implemented six different regression algorithms to reach a conclusion on the capabilities of different ML algorithms. The experimental results can vary significantly and provided a result of 91.51%, 91.62%, 91.72%, 91.86%, 85.70% and 95.38%, respectively. The Random Forest model achieved the top score in terms of Test R2, Test MAE, Test RMSE and CV R2. This study is very significant as it is the one of the first study that based on Bangladeshi crime data. Police department of Bangladesh is expected to get important insights from this research; like how crime rate can be predicted using ML. Police department of other countries may also get insight from this study. In future, I aim to implement the models in a device that'll assists police in predicting crime rate.