Stock market trading has evolved over years from human intervention to automated bots. This gives a huge wave of opportunities for data scientists to analyze and predict the stock price using various machine learning and deep learning techniques. There is a trade off between the traditional techniques and deep learning techniques for accuracy and performance in time series related problems. The techniques used as part of this thesis include Long Short-Term Memory (LSTM) and Autoregressive Integrated Moving Average (ARIMA) for forecasting analytics as an attempt to validate the combination of deep learning and machine learning techniques. The biggest advantage LSTM provides over traditional machine learning technique is the ability to remember the longer sequences of input data with the memory. The major question is whether LSTM alone is sufficient enough to achieve better accuracy or with additional training would be necessary to improve the prediction. However, the thesis question is whether an ensemble of LSTM and ARIMA can achieve better accuracy. The objective also includes other external techniques such as Natural Language Processing (NLP) to perform sentiment analysis and validate the final results. The end results show that with additional training and considering the ensemble of networks can provide better results. It is also observed that including NLP can help in understanding the domain, public response and improve the prediction
