Make yourself comfortable: Nudging urban heat and noise mitigation with smartwatch-based Just-in-time Adaptive Interventions (JITAI)

Public repository for the paper titled Make yourself comfortable: Nudging urban heat and noise mitigation with smartwatch-based Just-in-time Adaptive Interventions (JITAI) using Cozie-Apple data collected during the orenth and usk experiments. In the paper, orenth is referred to as 'Phase 1', and usk is referred to as 'Phase 2'.

Requirements

Install the required libraries:

python install -r requirements.txt

Organization

The repository is organized in three main directories:

1. 1_data
   This directory contains the source data files.
2. 2_processing
   This directory contains the notebooks for data processing and the output files of these notebooks.
3. 3_visualizations
   This directory contains all notebooks to create the data-based visualization of paper. The visualizations themselves are in a subdirectory named img.

Data

The data in this repository consists of the data collected during the orenth and usk experiments. The experiments lasted from October 2022 to July 2023. There are four main source data files in the [./1_data/](1_data directory):

• Onboarding survey answers: 1_onboarding_survey_all_participants.csv
• Weekly survey answers: 2_weekly_survey_data_all_participants.csv
• JITAI messages sent in phase 1: 3_jitai_data_participants_phase1_orenth.csv
• JITAI messages sent in phase 2: 4_to_be_published/1_data/4_jitai_data_participants_phase2_usk.csv

Column names JITAI data

Field name	Type	Unit	Description
action	String	-	Action taken by AWS Lambda function for JITAI messages
id_message	String	-	ID of message sent or not sent to participant
id_sender	String	-	Sender of message
recipients	Integer	-	Number of recipients of message
response_code	Integer	-	HTTP response code of OneSignal push notification API
timestamp_lambda	String	-	Timestamp of when AWS Lambda function for JITAI messages is executed

Values for action column

Values in action column	Description
no notification, conditions for message id not met.	No push notification is sent as none of the conditions are met.
send notification	Push notification is sent.
no action, duration threshold not reached. (MIN_JITAI_MESSAGE_INTERVAL)	Conditions to send a push notification are met, but no push notification is sent as there was already a push notification sent shortly before this one to the same participant.
no action, duration threshold not reached for message id ('send_interval_min').	Conditions to send a push notification are met, but no push notification is sent as there was already a push notification sent with the same id_message shortly before.
no action, max message threshold ('send_count_max' for message id) reached.	Conditions to send a push notification are met, but no push notification is sent as the daily limit for the number of push notifications with the same id_message was reached.
no action, max message threshold (all jitai) reached. (MAX_JITAI_MESSAGES_PER_DAY)	Conditions to send a push notification are met, but no push notification is sent as the daily limit for the number of push notifications over all was reached.
virtually fire notification	A push notification would have been sent if the debug mode had been disabled.

Values for id_message column

Values id_message column	Description
half_way	Inform participant that they have completed 50% of the experiment.
jitai_noise	JITAI message using a noise threshold value.
jitai_temperature	JITAI message using an outdoor air temperature threshold.
jitai_temperature_adaptive	JITAI message using personalized temperature threshold.
jitai_noise_adaptive	JITAI message using personalized noise threshold.

Values for recipients column

Values recipients column	Description
-1	No recipients of a push notification because no message was sent.
0	No recipients of a push notification despite the attempt to send a push notification.
1	One recipient of a push notification.

Processing

The personality scores are computed in 1_personality_score_calculation.ipynb. The scores are then saved in 2_all_personality_scores.csv. The scores are also merged with the weekly survey responses and saved in 3_weekly_survey_with_personality.csv.

Further, there are two notebooks for minor data processing:

• Labelling missing responses (Output file: 5_weekly_survey_with_personality_no_nan.csv)
• Removing redundant respones (Output file: 7_weekly_survey_with_personality_no_nan_one_response_per_week.csv)

Visualizations

The notebooks to create the data-based figures can be found in the visualizations directory. The resulting images can be found in the img sub-directory.

Figure	Notebook
Figure 7	figure_7_8_10.ipynb
Figure 8	figure_7_8_10.ipynb
Figure 9	figure_9.ipynb
Figure 10	figure_7_8_10.ipynb
Figure 12a	figure_12_jitai_changes.ipynb
Figure 12b	figure_12_jitai_changes.ipynb