This research paper is aimed at a practical analysis of the impact of the electroencephalogram (EEG) signal processing techniques on novel classification algorithms. The study examines how newly developed classification algorithms may perform significantly depending on the methods used to interpret EEG signals (brain activity data). In this context, first, we have conducted a review on recently contributed research articles based on EEG signal classification. Using this review, we are trying to recover the essential datasets, features, feature selection techniques, classification techniques, and applications. Next, the work is extended to study the nature of EEG data and its influence on data scaling. Further, we have also investigated the performance influence of feature selection technique on scaled data. Additionally, the convolutional neural network has been trained to learn from scaled and non-scaled data. Using four different experimental scenarios we have evaluated the performance of implemented data scaling and feature selection techniques over the EEG signal classification. Based on experimental analysis we have found that standardization is much beneficial than normalization. Additionally, it also helps to improve training performance of a classification model. However, with the merging of principal component analysis and both the scaling technique, we found the PCA and standardization are providing better training yield, but at the time of validation the standardized data is superior then other implemented techniques. Finally, conclusion and future extension plan have been discussed.