Data Science IOT Anomaly, Risk, and Failure Prediction Detection: Project Overview

• Any dataset can be implemented from similar IoT related devices that produce time series data.
• This project can help you to learn: **- how to approach sensors data **- how to find anomalies which can identify pumps or other IOT devices at risk, **- slopes in the time series record, another indicator of potential failure. **- python code which can be run on AWS Sagemaker or a desktop Jupyter notebook

Code and Resources Used

• Python Version: 3.7
• Packages/Libraries: pandas, numpy, matplotlib, pyplot, seaborn, sklearn, LabelEncoder, train_test_split, LogisticRegression, confusion_matrix, classification_report, accuracy_score
• Dataframe Shape
• Additional References
• [https://pythonsimplified.com/5-common-techniques-to-handle-imbalanced-data/](Techniques to Handle Imbalanced Data)
• [https://machinelearningmastery.com/failure-of-accuracy-for-imbalanced-class-distributions/](Failure of Accuracy)

IOT pump data for analytics Placed Data file located https://data.world/spectra-xenon/utility-sensor-data-set/