WiFi features by far seem to be the most used features, and they are producing great results.
\nThe question that arises is how much better can we make our results using WiFi features alone? What other features should we use and how should we use them?
If you haven’t already, I highly recommend checking out this github to understand the data and features: https://github.com/location-competition/indoor-location-competition-20
\nGenerally, the types of data in this competition can be categorized into 2 groups:
\n1) Accelerometer, Magnetic Field, Gyroscope, and Rotation Vector which follow a similar format (X axis, Y axis, Z axis)
\n2) WiFi and Beacons which also follow a similar format (identifier info, RSSI and strength info)
So which ones are useful?
\nWell logically, since WiFi is working so well, Beacons should perform similarly. Unfortunately, it sounds like people are not finding much success with it yet - read for yourself here https://www.kaggle.com/c/indoor-location-navigation/discussion/222488.
Another way features are being used is for trying to “predict the walk” of a path using the sensors. This is related more to the first group of features I mentioned above. For example, you can try to use Accelerometer to predict steps and Magnetic Field to predict direction, see here: https://www.kaggle.com/museas/estimate-the-walk-with-acce-and-magn. I see less use for the Gyroscope and Rotation Vector, but they may still be useful. This article gives a good explanation of what these sensors are: https://source.android.com/devices/sensors/sensor-types
\nHere are some useful resources for features:
\nVery commonly used WiFi features by BIZEN - https://www.kaggle.com/hiro5299834/indoor-navigation-and-location-wifi-features
\nIndoor XY_Floor features by Oscar Villarreal Escamilla - https://www.kaggle.com/oxzplvifi/indoor-xy-floor
\nLabel encoded wifi features by Jiwei Liu - https://www.kaggle.com/jiweiliu/wifi-label-encode
\nindoor-unified-wifi-ds by Kouki - https://www.kaggle.com/kokitanisaka/indoorunifiedwifids
I hope you found this useful and if you have questions or suggestions comment below.
", "messages": [ { "id": 1221247, "postDate": "2021-02-28T19:40:31.637Z", "content": "WiFi features by far seem to be the most used features, and they are producing great results.
\nThe question that arises is how much better can we make our results using WiFi features alone? What other features should we use and how should we use them?
If you haven’t already, I highly recommend checking out this github to understand the data and features: https://github.com/location-competition/indoor-location-competition-20
\nGenerally, the types of data in this competition can be categorized into 2 groups:
\n1) Accelerometer, Magnetic Field, Gyroscope, and Rotation Vector which follow a similar format (X axis, Y axis, Z axis)
\n2) WiFi and Beacons which also follow a similar format (identifier info, RSSI and strength info)
So which ones are useful?
\nWell logically, since WiFi is working so well, Beacons should perform similarly. Unfortunately, it sounds like people are not finding much success with it yet - read for yourself here https://www.kaggle.com/c/indoor-location-navigation/discussion/222488.
Another way features are being used is for trying to “predict the walk” of a path using the sensors. This is related more to the first group of features I mentioned above. For example, you can try to use Accelerometer to predict steps and Magnetic Field to predict direction, see here: https://www.kaggle.com/museas/estimate-the-walk-with-acce-and-magn. I see less use for the Gyroscope and Rotation Vector, but they may still be useful. This article gives a good explanation of what these sensors are: https://source.android.com/devices/sensors/sensor-types
\nHere are some useful resources for features:
\nVery commonly used WiFi features by BIZEN - https://www.kaggle.com/hiro5299834/indoor-navigation-and-location-wifi-features
\nIndoor XY_Floor features by Oscar Villarreal Escamilla - https://www.kaggle.com/oxzplvifi/indoor-xy-floor
\nLabel encoded wifi features by Jiwei Liu - https://www.kaggle.com/jiweiliu/wifi-label-encode
\nindoor-unified-wifi-ds by Kouki - https://www.kaggle.com/kokitanisaka/indoorunifiedwifids
I hope you found this useful and if you have questions or suggestions comment below.
", "rawMarkdown": "WiFi features by far seem to be the most used features, and they are producing great results.\nThe question that arises is how much better can we make our results using WiFi features alone? What other features should we use and how should we use them? \n\nIf you haven’t already, I highly recommend checking out this github to understand the data and features: [https://github.com/location-competition/indoor-location-competition-20](url) \nGenerally, the types of data in this competition can be categorized into 2 groups: \n1) Accelerometer, Magnetic Field, Gyroscope, and Rotation Vector which follow a similar format (X axis, Y axis, Z axis) \n2) WiFi and Beacons which also follow a similar format (identifier info, RSSI and strength info)\n\nSo which ones are useful?\nWell logically, since WiFi is working so well, Beacons should perform similarly. Unfortunately, it sounds like people are not finding much success with it yet - read for yourself here [https://www.kaggle.com/c/indoor-location-navigation/discussion/222488](url). \n\nAnother way features are being used is for trying to “predict the walk” of a path using the sensors. This is related more to the first group of features I mentioned above. For example, you can try to use Accelerometer to predict steps and Magnetic Field to predict direction, see here: [https://www.kaggle.com/museas/estimate-the-walk-with-acce-and-magn](url). I see less use for the Gyroscope and Rotation Vector, but they may still be useful. This article gives a good explanation of what these sensors are: [https://source.android.com/devices/sensors/sensor-types](url) \n\nHere are some useful resources for features:\nVery commonly used WiFi features by BIZEN - [https://www.kaggle.com/hiro5299834/indoor-navigation-and-location-wifi-features](url) \nIndoor XY_Floor features by Oscar Villarreal Escamilla - [https://www.kaggle.com/oxzplvifi/indoor-xy-floor](url) \nLabel encoded wifi features by Jiwei Liu - [https://www.kaggle.com/jiweiliu/wifi-label-encode](url)\nindoor-unified-wifi-ds by Kouki - [https://www.kaggle.com/kokitanisaka/indoorunifiedwifids](url) \n\nI hope you found this useful and if you have questions or suggestions comment below.", "votes": 19 } ], "comments": [], "raw_markdown_by_id": { "1221247": "WiFi features by far seem to be the most used features, and they are producing great results.\nThe question that arises is how much better can we make our results using WiFi features alone? What other features should we use and how should we use them? \n\nIf you haven’t already, I highly recommend checking out this github to understand the data and features: [https://github.com/location-competition/indoor-location-competition-20](url) \nGenerally, the types of data in this competition can be categorized into 2 groups: \n1) Accelerometer, Magnetic Field, Gyroscope, and Rotation Vector which follow a similar format (X axis, Y axis, Z axis) \n2) WiFi and Beacons which also follow a similar format (identifier info, RSSI and strength info)\n\nSo which ones are useful?\nWell logically, since WiFi is working so well, Beacons should perform similarly. Unfortunately, it sounds like people are not finding much success with it yet - read for yourself here [https://www.kaggle.com/c/indoor-location-navigation/discussion/222488](url). \n\nAnother way features are being used is for trying to “predict the walk” of a path using the sensors. This is related more to the first group of features I mentioned above. For example, you can try to use Accelerometer to predict steps and Magnetic Field to predict direction, see here: [https://www.kaggle.com/museas/estimate-the-walk-with-acce-and-magn](url). I see less use for the Gyroscope and Rotation Vector, but they may still be useful. This article gives a good explanation of what these sensors are: [https://source.android.com/devices/sensors/sensor-types](url) \n\nHere are some useful resources for features:\nVery commonly used WiFi features by BIZEN - [https://www.kaggle.com/hiro5299834/indoor-navigation-and-location-wifi-features](url) \nIndoor XY_Floor features by Oscar Villarreal Escamilla - [https://www.kaggle.com/oxzplvifi/indoor-xy-floor](url) \nLabel encoded wifi features by Jiwei Liu - [https://www.kaggle.com/jiweiliu/wifi-label-encode](url)\nindoor-unified-wifi-ds by Kouki - [https://www.kaggle.com/kokitanisaka/indoorunifiedwifids](url) \n\nI hope you found this useful and if you have questions or suggestions comment below." } }