Repository files navigation

• Introduction
• Python
• R
• InfluxDB

The Time Series Anomaly Detection repo contains several examples of anomaly detection algorithms for use with time series data sets.

Examples can be found in thepython directory andr directory.

InfluxDB and Grafana are optionally included in the Docker stack for data storage and visualization purposes. Telegraf has been included to fill the InfluxDB with dummy metric data.

Redis is optionally included in the Docker stack for caching (i.e. memoization) purposes.

If you're using Docker, executebuild.sh to get started.

Examples are typically written in python. From the.env.example file, you can see that scripts are written in python 3.8.2. A list of module dependencies can be found in theDockerfile andrequirements.txt. You aren't forced to use Docker, and can use something like Conda instead if that's your preference.

If you opt to use Docker, you can view theMakefile for relevant Docker commands. Themake penter command will create a new container and execute the python CLI. Themake prun command will run a python script. For example,make prun d=ts_price_anomaly_detection s=view will runts_price_anomaly_detection/view.py

Example anomaly detection algorithms can be found in thepython directory, and each example directory contains a similar structure. When exploring an example, you should first read the README.md and references.md files. The references.md file will provide you with a relevant link to a tutorial page and data set. Download the recommended data set and place it in the local data directory (don't place it in theroot data directory).

You can then execute various python scripts to analyze and model the data. It's recommended that you run explore.py then view.py first to better understand the distribution of the data.

Additional examples are written in R. From the.env.example file, you can see that R scripts are written in version 3.6.3. A list of additional R packages can be found in theDockerfile.

As thedocker-compose.yml file shows, this repo employs therocker/tidyverse image which already includes the tidyverse collection and RStudio server.

If you opt to use Docker, you can view theMakefile for relevant Docker commands. Themake renter command will allow users to execute shell commands within the R container. Themake rrun command will run an R script. For example,make rrun s=bitcoin_anomalies will run$R_STUDIO_USER/bitcoin_anomalies.r

Example anomaly detection algorithms can be found in ther directory. You can then execute various r scripts to analyze and model the data.

InfluxDB is a time series database. For those who are unfamiliar, more information can be found atinfluxdata.com. InfluxDB can be combined withGrafana to analyze and visualize the data. View the.env.example file to configure your InfluxDB & Grafana versions and ports.

CSV files can be easily imported to your InfluxDB instance using thecsv-to-influxdb package. Each example directory will contain a README.md file with acsv-to-influxdb.py command to execute.