Download

Our dataset is based on the thirteen users who collected the data during their daily activities. The dataset includes all sensors available in phones and distinguishes five transportation modes: being on a car, on a bus, on a train, standing still and walking.
The dataset was used to build different models, whit different classification algorithms (Decision Tree, Random Forest, Support Vector Machine and Neural Network) reaching a maximum level of accuracy of 96%.

---

Publications

If you use the dataset in one of your works, please cite the following publication:
Carpineti C., Lomonaco V., Bedogni L., Di Felice M., Bononi L., "Custom Dual Transportation Mode Detection by Smartphone Devices Exploiting Sensor Diversity", in Proceedings of the 14th Workshop on Context and Activity Modeling and Recognition (IEEE COMOREA 2018), Athens, Greece, March 19-23, 2018 [Pre-print available]

---

Principal dataset

The following datasets are those used in our tutorial section and in the code that you can find in our Github repository.

Raw Data

The raw data is those collected through the smartphone sensors of our volunteers. These data were not cleaned.
For more information, go here.

Dataset	Format	Download
Raw data of dataset's users	CSV

---

Dataset balanced

Balanced dataset contains features extracted from raw data. It was built with two different time windows (5 seconds and a half second).
For more information, go here.

#	Dateset	Format	Download
1	Dateset balanced (5 seconds window)	CSV JSON
2	Dataset balanced (0.5 second window)	CSV JSON

---

Custom dataset

The following datasets were generated by extracting the features of certain categories (eg dataset for the Walking action, student datasets, user datasets over / under 30 years, etc ...) from the balanced dataset.

Custom dataset - categories

Balanced dataset that only contains the features of the selected categoryc (bus, car, train, still, walking).

#	Dateset	Format	Categories	Download
1	Dateset 5 seconds window	CSV	Bus Car Still Train walking
2	Dateset half seconds window	CSV	Bus Car Still Train walking

---

Custom dataset - sensor

Balanced dataset that only contains the features of the selected sensor.

#	Dataset	Format	Sensor	Download
1	Dataset 5 second window	CSV	Accelerometer Game rotation vector Gravity Gyroscope Gyroscope uncalibrated Light Linear acceleration Magnetic field Magnetic field uncalibrated Orientation Pressure Proximity Rotation vector Sound Speed Step counter
2	Dataset half second window	CSV	Accelerometer Game rotation vector Gravity Gyroscope Gyroscope uncalibrated Light Linear acceleration Magnetic field Magnetic field uncalibrated Orientation Pressure Proximity Rotation vector Sound Speed Step counter

---

Custom dataset - user under or over 30 years

Balanced dataset that only contains the features of the selected category of people:

• people under 30 years old;
• people over 30 years old.

#	Dataset	Format	Download
1	Dataset 5 second window	CSV	Under 30 years Over 30 years
1	Dataset 5 second window	CSV	Under 30 years Over 30 years

---

Custom dataset - user that are student or not

Balanced dataset that only contains the features of the selected category of people:

• people who are students;
• people who are not students (do other jobs).

#	Dataset	Format	Download
1	Dataset 5 second window	CSV	Student Not Student (other jobs)
1	Dataset 5 second window	CSV	Student Not Student (other jobs)

---