dotnet / dotnet/machinelearning

[AnomalyDetection] ExpectedRange higher than expected?

Open
#5,630 1 comment 0 reactions 0 assignees View on GitHub
area-TimeSeries Priority:2
Dominant language
C#
Stars
9.4k
Forks
2k
Avg merge
2d 20h
Merged PRs (30d)
11

Description

### System information

- **OS version/distro**:
- **.NET Version (eg., dotnet --info)**: master

### Issue

- **What did you do?** I apply anomaly detection with csv file below
- **What happened?** Expected the expectedRange around 30K-40K, yet it's >50K.
- **What did you expect?** same above
![image](https://user-images.githubusercontent.com/3010893/109183600-f95b7700-7742-11eb-9338-9563e377a9d7.png)

**Please assign the issue** to: https://github.com/guinao and https://github.com/mengaims

### Source code / logs
```
var anomalyDetectorOptions = new SrCnnEntireAnomalyDetectorOptions()
{
DetectMode = //WithMargin,
Sensitivity = 84,
DeseasonalityMode = SrCnnDeseasonalityMode.Median,
Period = -1,
Threshold =0.12,
};

anomalyResult = this.mlContext.AnomalyDetection.DetectEntireAnomalyBySrCnn(
...
anomalyDetectorOptions);
```

Input data:
Value,Date
39800,2020-01-01 00:00:00
35000,2020-02-01 00:00:00
35600,2020-03-01 00:00:00
31700,2020-04-01 00:00:00
34900,2020-05-01 00:00:00
35200,2020-06-01 00:00:00
38900,2020-07-01 00:00:00
63981,2020-08-01 00:00:00
42000,2020-09-01 00:00:00
39400,2020-10-01 00:00:00
33700,2020-11-01 00:00:00
37800,2020-12-01 00:00:00
34200,2021-01-01 00:00:00

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.