Threshold Anomaly Detection

This algorithm detects out of bounds data-points from a time-series dataset and sets them to 0.

Table of Contents

Introduction
Inputs
Outputs
Algorithm Console

Introduction

By using the pre-defined upper & lower bounds, this algorithm will keep all data-points that lie within the thresholds. All data-points that lie outside are set to 0.

Inputs

This algorithm has two input formats; JSON Formatting and Array Formatting.

JSON Format

{  
   "data": Double[],
   "lowerBound": Double,
   "upperBound": Double
}

{  
   "data": [1,2,3,4,3,4,6,7,5,4,2,1,1],
   "lowerBound": 1.4,
   "upperBound":  6.5
}

data - (required) - The time series input data you wish to detect anomalies on, this algorithm assumes the values are evenly spaced. (dx(1) == dx(n))
lowerBound - (optional) - The lower boundary limit, the default value is 0.
upperBound - (optional) - The upper boundary limit, the default value is 0.

Array Format

[
   data: Double[],
   lowerBound: Double,
   upperBound:  Double
]

[[1,2,3,4,3,4,6,7,5,4,2,1,1], 1.4, 6.5]

data - (required) - The time-series input data you wish to detect anomalies on, this algorithm assumes the values are evenly spaced. (dx(1) == dx(n))
lowerBound - (required) - The lower boundary limit.
upperBound - (required) - The upper boundary limit.

Outputs

This algorithm has json output for json input, and array output for array input.

JSON Format

{  
   "output": Double[]
}

{  
   "output": [0,2,3,4,3,4,6,0,5,4,2,0,0]
}

output - The filtered dataset, if a data-point was considered out of bounds it’s value is set to 0.

Array Format

output: Double[]

[0,2,3,4,3,4,6,0,5,4,2,0,0]

output - The filtered dataset, if a datapoint was considered out of bounds it’s value is set to 0.

Sign In

Threshold Anomaly Detection

Run an Example

Install & Use