Dyson - sporedata/researchdesigneR GitHub Wiki

General description

Dyson indoor air quality is a unique dataset with around 6 million indoor air quality sensors measuring particulate matter (PM2.5 and PM10), volatile organic compounds, and nitrogen dioxide. Dyson indoor air quality sensor data have high resolution information about the US, Singapore, and Australia.

Factors to consider when using database (for research)

As the sensors are unevenly distributed throughout the country, they should be combined into a regular spatiotemporal data cube. An approach to this aggregation uses a k-nearest neighbors (kNN) model. kNN builds the model using raw data points and outputs a fine-grained regular grid in which each point is the average of the k closest raw points. Larger k values output a smoother model but also hides local peaks. A Gaussian kernel also can be used to compute the distances.

Use cases and companion methods

  • Can be used to determine indoor exposure to air quality of different levels.

Variable categories

  • Data quality metrics (e.g., absolute completeness, sensor error loss, practical completeness)
  • Air quality metrics (e.g., 95th percentile exposure, number and duration of exceedances of recommended levels, number and duration of exposure events)
  • Air compounds (e.g., PM10 and PM2.5 concentrations, volatile organic compounds and NO2 index)

Limitations

  • Data points can be sparse in areas where Dyson purifiers are not present.
  • The validation of additional air quality metrics is being conducted over time, and so might not be available.
  • Since Dyson is constantly improving the quality of its air sensors, the measurement characteristics obtained from the latest generation of its air purifiers might have different metrics when compared to earlier generations.
  • When purifiers are not turned on for whatever reason (power failure, structural changes in a building, etc), the longitudinal data capture will demonstrate missing values.

Data access

Data access is restricted to those establishing collaboration agreements with Dyson.

SporeData data dictionaries

  • sddatadictionaries/Dyson_air_and_data_quality_metrics.pdf