The objective in this project will be to detect anomalies, which correspond to cyber-attacks, in a time-series dataset.

Training dataset only consists of nominal instances, so your goal in training will be to develop an algorithm which can learn some nominal patterns and detect deviations from the nominal patterns (i.e., anomalous instances) in the testing dataset. Testing dataset will contain both nominal and anomalous instances. Your performance will be measured in terms of two types of error: number of false positives (i.e., false alarms) and number of false negatives (i.e., missed anomalies). You can use ROC curve, AUC value, and F1 value to measure your performance. After your final submission, your code will be run to confirm your results. Note that you are expected to do a comprehensive analysis by using multiple algorithms. You are not restricted to the algorithms studied in class.

For the evaluation, please download the sample file in csv format. The left column is for the timestamps and the right column is for the predicted labels. Please replace the right column by your predicted labels (0 for Normal/ 1 for Attack) and upload the file named as
"project_anomaly_fullname"