K-Means_Influx

Using K-Means for anomaly detection for time series data with InfluxDB and Chronograf

Data:

EKG Data from: https://www.kaggle.com/ecerulm/apneaecg

Please view the following blogs for more info on this repo:

https://www.influxdata.com/blog/why-use-k-means-for-time-series-data-part-one/ https://www.influxdata.com/blog/why-use-k-means-for-time-series-data-part-two/ https://www.influxdata.com/blog/why-use-k-means-for-time-series-data-part-three/

Files:

EKG Data with anomaly in line protocol (data ingest format for InfluxDB):

anomaly.txt

EKG Data without anomaly in line protocol (data ingest format for InfluxDB):

norm.text

Use Kmeans to reconstruct the data, set threshold, detect anomalies, write error to InfluxDB with python CL:

Anomaly_Detection.ipynb