Precise reporting from imprecise data sources

Data Reliability Landscape

Telemetry, a 19th Century concept with 21st Century application continues to expand its horizons with technology enablement. Expectations on reliable data and decision making based on data have increased but the environments in which telemetric devices operate remain as hostile as ever. In applications where reports are compiled from time series data emanating from telemetric devices, erroneous or missing data points undermine the accuracy and usage of these reports.

Introducing Ovidius Smart Data, Sidero has harnessed machine learning techniques to auto-correct and compensate for erroneous or missing data points, providing the means for reliable and accurate reports.

Ovidius Smart Data

What is it?

  • Ovidius Smart Data is a software tool which deploys in AWS and uses underlying AWS AI and ML services
  • It takes in raw data files from devices and identifies anomalous data samples and missing data points
  • Using machine learning and artificial intelligence it auto-corrects the bad data samples and generates a modified data file which can be further post processed for reporting or decision making

Benefits

  • Eliminates the need for time consuming and error prone manual correction of data by factors of several hundred
  • Enhances the accuracy and reliability of data emanating from telemetric devices operating in environmentally hostile areas
  • Very cost efficient to operate due to its bursty compute needs optimised for public cloud deployment

Use cases

  • In situations where accurate reporting or decision making based on time series generated data is required
  • In situations where data from telemetric data sources can be compromised by challenging electrical or physical environmental characteristics


How does it work?

Data is segmented into component parts and uploaded securely into AWS S3 buckets and mapped with AWS SageMaker. Using ARIMA and linear regression model, time series data flow/ velocity is analysed to identify anomalies and to predict correct values based on previously observed values. Predicted values are checked with error rate to evaluate the model performance for accuracy of data. The predicted data sets are converted into CSV files and stored in S3 buckets to upload to AWS QuickSight for data visualisation and presentation.

DevSecOps With Machine Learning



For more information contact us at
info@sidero.ie