Multivariate Gait Data

Donated on 12/15/2022

Bilateral (left, right) joint angle (ankle, knee, hip) times series data collected from 10 healthy subjects under 3 walking conditions (unbraced, knee braced, ankle braced). For each condition, each subject’s data consists of 10 consecutive gait cycles.

Dataset Characteristics

Sequential, Multivariate, Time-Series

Subject Area

Life Science

Associated Tasks

Classification, Regression, Clustering

Attribute Type

Real, Categorical, Integer

# Instances


# Attributes



For what purpose was the dataset created?

Biomechanical analysis of human locomotion

Who funded the creation of the dataset?

National Science Foundation (#0540834) and Mary Jane Neer Disability Research Fund at the University of Illinois

Additional Information

This dataset is a six dimensional array of joint angle data: 10 subjects x 3 conditions x 10 replications x 2 legs x 3 joints x 101 time points. The data were recored from ten subjects under three different conditions: normal (unbraced) walking on a treadmill, walking on a treadmill with a knee-brace on the right knee, and walking on a treadmill with an ankle brace on the right ankle. For each subject in each condition, ten consecutive gait cycles (replications) are included, where each gait cycle starts and ends at heel-strike. For each gait cycle, the data were normalized to consist of 101 time points representing 0%,…,100% of the gait cycle. Six joint angles are included, which comprise all combinations of leg (left and right) and joint (ankle, knee, hip). The data were collected at the Human Dynamics and Controls Laboratory at the University of Illinois at Urbana-Champaign. Details of the experimental setup can be found in Shorter et al. (2008). Details on the data preprocessing can be found in Helwig et al. (2011). The data were published as supplementary materials by Helwig et al. (2016). Attribute Information: 1. subject: 1 = subject 1, …, 10 = subject 10 (integer) 2. condition: 1 = unbraced, 2 = knee brace, 3 = ankle brace (integer) 3. replication: 1 = replication 1, …, 10 = replication 10 (integer) 4. leg: 1 = left, 2 = right (integer) 5. joint: 1 = ankle, 2 = knee, 3 = hip (integer) 6. time: 0 = 0% gait cycle, …, 100 = 100% gait cycle (integer) 7. angle: joint angle in degrees (real valued)

Citation Requests/Acknowledgements

Any use of these data should cite the following three papers and include a statement along the lines of “The data were collected by Shorter et al. (2008), preprocessed by Helwig et al. (2011), and published by Helwig et al. (2016).” Shorter, K. A., Polk, J. D., Rosengren, K. S., Hsiao-Wecksler, E. T. (2008). A new approach to detecting asymmetries in gait. Clinical Biomechanics. 23(4), 459-467. Helwig, N. E., Hong, S., Hsiao-Wecksler E. T., & Polk, J. D. (2011). Methods to temporally align gait cycle data. Journal of Biomechanics, 44(3), 561-566. Helwig, N. E., Shorter, K. A., Ma, P. & Hsiao-Wecksler, E. T. (2016). Smoothing spline analysis of variance models: A new tool for the analysis of cyclic biomechanical data. Journal of Biomechanics, 49(14), 3216-3222.


Attribute NameRoleTypeDescriptionUnitsMissing Values
"subject"FeatureCategorical1 = subject 1, …, 10 = subject 10false
"condition"FeatureCategorical1 = unbraced, 2 = knee brace, 3 = ankle bracefalse
"replication"FeatureNumerical - Discrete1 = replication 1, …, 10 = replication 10false
"leg"FeatureCategorical1 = left, 2 = rightfalse
"joint"FeatureCategorical1 = ankle, 2 = knee, 3 = hipfalse
"time"FeatureNumerical - Discrete0 = 0% gait cycle, …, 100 = 100% gait cyclefalse
"angle"FeatureNumerical - Continuousjoint angle in degreesfalse

1 to 10 of 7

Introduction Paper


1 citations


Classificationclinical medicinehierarchical time seriesMulti-class classification Multivariate regressionsensor datatime serieswearable sensing


Nathaniel Helwig

University of Minnesota

Elizabeth Hsiao-Wecksler

University of Illinois at Urbana-Champaign


This dataset is licensed under a Creative Commons Attribution 4.0 International (CC BY 4.0) license.

This allows for the sharing and adaptation of the datasets for any purpose, provided that the appropriate credit is given.

By using the UCI Machine Learning Repository, you acknowledge and accept the cookies and privacy practices used by the UCI Machine Learning Repository.

Learn More