The paper presented here describes the predicted survival time for brain tumor patients using Magnetic Resonance Imaging (MRI). The accuracy is improved using the denoising wavelet transform (DWT) method. For this work BraTS, a dataset is used. MRI images are used to extract the histogram features so that the prediction model can be trained using the machine learning methods. MRI information is damaged due to the noise in MRI imaging. And the 2D wavelet transform was able to improve the results. The SVM with a 10 folds cross-validation helps to achieve the best accuracy by Daubechies 4 level 4 (db4-L4). With the same 10 folds, Daubechies 2 level 1 and 3 produces better results when the age factor is removed. An accuracy of 66.7% is achieved with a 10 % hold validation method in Daubechies 2 level 3.