Here I detected anomaly in stock prediction. After some simple data preprocessing, I build a LSTM AutoEncoder. After training it for 15 epochs I got a loss of 0.0278 on val set. Then I plotted the result i.e. val loss against epochs. Finally I plotted the result from my trained model, which almost detect the anomaly perfectly.
The source code in python is available in the notebook. The dataset is also added.
Original dataset can also be found here