Combine Dataframes

General Explanation

The Combining DataFrames Pipeline is designed to preprocess and merge raw accelerometer data and self-reports into a unified dataset. It simplifies the process of aligning timestamps, aggregating movement data within a specified time window, and dynamically adding contextual labels from self-reports. This pipeline is modular, making it reusable and adaptable for different stages of analysis, from preprocessing to model training.

Click to view diagram

Use Case

Combine raw accelerometer and self-reports data: Create a structured dataset ready for feature extraction or model training. Preprocess data consistently: Ensure uniform handling of timestamps, labels, and accelerometer data across different datasets. Prepare data for exploratory or predictive analysis: Use the resulting combined dataset as the foundation for understanding relationships between movement patterns and self-reported labels.

Inputs

Input Type	Description
Raw Accelerometer Data	CSV file with columns for x, y, z axes, timestamps (timeOfNotification), and participantId.
Self-Reports Data	CSV file with columns for timeOfNotification, participantId, and emotion labels (e.g., arousal, valence).
Configuration Settings	- time_window: Specifies the size (in minutes) of the time window for aggregating accelerometer data around each self-report.
	- label_columns: A list of columns from the self-reports dataset to use as labels (e.g., ["arousal", "valence"]).

Outputs

It Outputs a Dataset with these Columns:

Column Name	Description
participantId	Participant identifier.
selfreport_time	Timestamp of the self-report.
accel_time	Timestamp of the accelerometer reading.
x	Accelerometer reading for the x axis.
y	Accelerometer reading for the y axis.
z	Accelerometer reading for the z axis.
Emotion labels	Emotion labels from label_columns.
groupid	A unique identifier for each group of self-report and its corresponding accelerometer data.