Introduction to Data Analysis
Competition: Clustering
Overview
§ The goal is to improve your model’s performance as much as possible.
§ You can improve your model’s performance through data preprocessing, model selecti on, and hyperparameter tuning.
§ You can submit your model’s results to Kaggle and view your scores in real-time on the le aderboards.
• You can only submit a maximum of 20 per day.
Rules
§ In this competition, you can build a model using any clustering method.
§ You must participate as an individual team in this competition.
Evaluation
§ Submissions are evaluated on the Adjusted Rand Index between the ground truth clust er labels of the data and your predicted cluster labels (assignment).
• Please refer to the link.
§ You are not given the number of ground truth clusters or any training labels.
• This is a completely unsupervised problem.
• We recommend cluster indices that start with 0. (e.g., 0, 1, 2, ...)
Dataset
Health Condition levels Dataset
• You are given data with features such as demographic features and lifestyle features to cluster people’s health condition levels.
Dataset Description
Feature Name Feature Type
Lifestyle_feature1 Categorical
Lifestyle_feature2 Numerical
Lifestyle_feature3 Categorical
Liefstyle_feature4 Numerical
Height Numerical
Smoker Categorical
Medical_history Categorical
Weight Numerical
Age Numerical
Sex Categorical
DFC Numerical
SC Categorical
EF Numerical
FI Numerical
AC Categorical
EDUT Numerical
HSM Categorical
Dataset Description
Abbreviations
• Dietary Fiber Consumption (DFC)
• Snack Consumption (SC)
• Exercise Frequency (EF)
• Fluid Intake (FI)
• Alcohol Consumption (AC)
• Electronic Device Usage Time (EDUT)
• Health Signal Monitoring (HSM)
Competition
§ You will enter the competition individually through Kaggle.
§ Kaggle Site: https://www.kaggle.com
§ Competition URL: https://www.kaggle.com/t/1459d64d96ce491e9b7aa42c40b6fa66
Competition
Kaggle-site
① Overview: Description of the task
② Data: Files to be used in the competition
③ Leaderboard: Current score updated in real time
④ Rules: What you must follow in the competition
⑤ My submissions: Files you submitted
⑥ Submit Predictions: Submission
