First, we need data

Multiple labeled time series datasets exist for anomaly detection. We provide below a detailed list of these datasets, which are included in TSB-UAD benchmarks and can be downloaded here.

Datasets Descriptions

Dataset	Description
Dodgers	is a loop sensor data for the Glendale on-ramp for the 101 North freeway in Los Angeles and the anomalies represent unusual traffic after a Dodgers game.
ECG	is a standard electrocardiogram dataset and the anomalies represent ventricular premature contractions. We split one long series (MBA_ECG14046) with length ∼ 1e7) to 47 series by first identifying the periodicity of the signal.
IOPS	is a dataset with performance indicators that reflect the scale, quality of web services, and health status of a machine.
KDD21	is a composite dataset released in a recent SIGKDD 2021 competition with 250 time series.
MGAB	is composed of Mackey-Glass time series with non-trivial anomalies. Mackey-Glass time series exhibit chaotic behavior that is difficult for the human eye to distinguish.
NAB	is composed of labeled real-world and artificial time series including AWS server metrics, online advertisement clicking rates, real time traffic data, and a collection of Twitter mentions of large publicly-traded companies.
NASA-SMAP and NASA-MSL	are two real spacecraft telemetry data with anomalies from Soil Moisture Active Passive (SMAP) satellite and Curiosity Rover on Mars (MSL). We only keep the first data dimension that presents the continuous data, and we omit the remaining dimensions with binary data.
SensorScope	is a collection of environmental data, such as temperature, humidity, and solar radiation, collected from a typical tiered sensor measurement system.
YAHOO	is a dataset published by Yahoo labs consisting of real and synthetic time series based on the real production traffic to some of the Yahoo production systems.
Daphnet	contains the annotated readings of 3 acceleration sensors at the hip and leg of Parkinson’s disease patients that experience freezing of gait (FoG) during walking tasks.
GHL	is a Gasoil Heating Loop Dataset and contains the status of 3 reservoirs such as the temperature and level. Anomalies indicate changes in max temperature or pump frequency.
Genesis	is a portable pick-and-place demonstrator which uses an air tank to supply all the gripping and storage units.
MITDB	contains 48 half-hour excerpts of two-channel ambulatory ECG recordings, obtained from 47 subjects studied by the BIH Arrhythmia Laboratory between 1975 and 1979.
OPPORTUNITY (OPP)	is a dataset devised to benchmark human activity recognition algorithms (e.g., classiffication, automatic data segmentation, sensor fusion, and feature extraction). The dataset comprises the readings of motion sensors recorded while users executed typical daily activities.
Occupancy	contains experimental data used for binary classiffication (room occupancy) from temperature, humidity, light, and CO2. Ground-truth occupancy was obtained from time stamped pictures that were taken every minute.
SMD (Server Machine Dataset)	is a 5-week-long dataset collected from a large Internet company. This dataset contains 3 groups of entities from 28 different machines.
SVDB	includes 78 half-hour ECG recordings chosen to supplement the examples of supraventricular arrhythmias in the MIT-BIH Arrhythmia Database.

Datasets characteristics

The following table summarizes different characteristics of the datasets.

Dataset	Count	Avg length	Avg number of anomalies	Avg number of abnormal points
Dodger	1	50400.0	133.0	5612
ECG	53	230351.9	195.6	15634
IOPS	58	102119.2	46.5	2312.3
KDD21	250	77415.06	1.0	196.5
MGAB	10	100000.0	10.0	200.0
NAB	58	6301.7	2.0	575.5
SensorScope	23	27038.4	11.2	6110.4
YAHOO	367	1561.2	5.9	10.7
NASA-MSL	27	2730.7	1.33	286.3
NASA-SMAP	54	8066.0	1.26	1032.4
Daphnet	45	21760.0	7.6	2841.0
GHL	126	200001.0	1.2	388.8
Genesis	6	16220.0	3.0	50.0
MITDB	32	650000.0	210.1	72334.3
OPP	465	31616.9	2.0	1267.3
Occupancy	10	5725.8	18.3	1414.5
SMD	281	25562.3	10.4	900.2
SVDB	115	230400.0	208.0	27144.5

You may find more details (and the references) in this paper.