In this research, we studied the impact of the 5'UTR sequences on translation. This is done by generating various features describing the 5'UTR. Those features are then used as input in regression and classification models, that have the Translation Initiation Rates as target. The reason why it is of such importance, is dual. Because is very useful to be able to predict the initiation rates for new sequences and consequently be able to synthesize new sequences with high initiation rate. Additionally, it is important to understand which of the elements, located in the 5'UTR influence the translation initiation rates and thus the translation. Aim of this research is detecting those features from the 5'UTR of yeast's mRNA, that lead to higher translation initiation rates. In order to achieve this, data mining and machine learning techniques were used to build predictive models. To sum up, we have shown that predicting the exact initiation rates originating from stochastic models is not a trivial task and that many features are of no to little importance. Classification between low and high initiation rates seems to be more doable but there is still space for improvement and future research.