Collection of training data is performed with the help of IOT devices since the goal is to develop a AI based smoke detector device.

The dataset features in detail:

Air Temperature
Air Humidity
TVOC: Total Volatile Organic Compounds; measured in parts per billion (Source)
eCO2: co2 equivalent concentration; calculated from different values like TVCO
Raw H2: raw molecular hydrogen; not compensated (Bias, temperature, etc.)
Raw Ethanol: raw ethanol gas (Source)
Air Pressure
PM 1.0 and PM 2.5: particulate matter size < 1.0 µm (PM1.0). 1.0 µm < 2.5 µm (PM2.5)
Fire Alarm: ground truth is "1" if a fire is there
CNT: Sample counter
UTC: Timestamp UTC seconds
NC0.5/NC1.0 and NC2.5: Number concentration of particulate matter. This differs from PM because NC gives the actual number of particles in the air. The raw NC is also classified by the particle size: < 0.5 µm (NC0.5); 0.5 µm < 1.0 µm (NC1.0); 1.0 µm < 2.5 µm (NC2.5);

The sample rate is 1Hz for all sensors. To keep track of the data, a UTC timestamp is added to every sensor reading.

Sample submission: You should submit a CSV file with a header row and the sample submission can be found below:

Fire Alarm

0

1

0

1

1

Etc.

Note that the header name should be ‘Fire Alarm’ else it will throw an evaluation error.

Steps to follow to submit your notebook:

Go to the Notebook section on this challenge page
Click on the 'Upload Notebook' button on top left corner as shown in the image below:
Make sure to make the visibility of your notebook as 'public' while uploading the notebook.
Provide Notebook Title, Summary, add some tags, go to next, chose a file and click on Upload.

Data Sprint 89 - Sensor-Fusion Smoke Detection Classification