Self-mutilation and suicide are negative consequences of depression disorder, if it is left untreated. Two commonly prescribed ways of treatments for depression are Selective serotonin reuptake inhibitors (SSRIs) and repetitive transcranial magnetic stimulation (rTMS) therapies. Although the effectiveness of these two therapies have been approved by food and drug administration (FDA), the successful rate to these therapies is around 50%. In general, a psychiatric prescribes one of the therapies based on his/her experience and waits for a period of time to check the improvement. However, if the therapy fails to reduce depression levels, the risk of self-harm or suicide may increase. This paper proposes a robust software called Therapy Outcomes Predictor by Electroencephalogram (TOP-EEG) to predict the improvement of two different therapies for depressed patients based on pretreatment electroencephalogram (EEG) signals. The TOP-EEG software utilizes a novel recently proposed signal processing technique called dynamic graph mode decomposition (DGMD) to automatically decompose the EEG signals into the intrinsic mode functions (IMFs). Then, five entropy-based features quantize the value of the complexity and randomness of the IMFs. Statistically significant features are selected using the Kruskal-Wallis test and fed into traditional machine learning algorithms and artificial neural network architectures for classification. A 10-fold cross-validation strategy is applied during training and testing to minimize bias in the results. The cascade-forward neural network (CFNN) architecture shows the best performance among the other classification algorithms. The TOP-EEG software is evaluated using two different databases: SSRI and rTMS therapies, including data from 30 and 15 depressed patients, respectively. The results show that the TOP-EEG software achieves classification accuracy levels of 93.16% and 94.59% to predict the outcomes of the SSRI and rTMS therapies, respectively. This is the first time in the literature a software is developed to predict the outcomes of two therapies for depression and recommend the best course of treatment. The software is reliable and can be used in clinics and hospitals to assist neurologists and psychiatrists in prescribing the most effective treatment for depression.