Datasets for Chapter 6 Experiments

Sources from TsFile VLDB paper (Table 1).

Dataset Profile

Quick Start

# 1. Download raw data (see below for URLs)
# 2. Prepare each dataset
python3 prepare_redd.py    --raw-dir ./raw/redd    --out-dir ./prepared/redd
python3 prepare_geolife.py --raw-dir ./raw/geolife --out-dir ./prepared/geolife
python3 prepare_tdrive.py  --raw-dir ./raw/tdrive  --out-dir ./prepared/tdrive
bash    prepare_tsbs.sh     --out-dir ./prepared/tsbs

# 3. Or prepare all at once
bash prepare_all.sh

Data Sources

REDD (Reference Energy Disaggregation Data Set)

• Paper: [Kolter & Johnson 2011]
• URL: http://redd.csail.mit.edu/ (requires registration)
• Download low_freq.tar.bz2, extract to raw/redd/
• Expected structure: raw/redd/house_{1..6}/channel_{1..N}.dat

GeoLife

• Paper: [Zheng et al. 2010]
• URL: https://download.microsoft.com/download/F/4/8/F4894AA5-FDBC-481E-9285-D5F8C4C4F039/Geolife%20Trajectories%201.3.zip
• Extract to raw/geolife/
• Expected structure: raw/geolife/Data/{000..181}/Trajectory/*.plt

TDrive

• Paper: [Yuan et al. 2010, 2011]
• URL: https://www.microsoft.com/en-us/research/publication/t-drive-driving-directions-based-on-taxi-trajectories/
• Download both parts, extract to raw/tdrive/
• Expected structure: raw/tdrive/{1..10357}.txt

TSBS (Time Series Benchmark Suite)

• Repo: https://github.com/timescale/tsbs
• IoT use case (trucks with coordinates, velocity)
• Generated via tsbs_generate_data, no manual download needed
• Requires Go toolchain

Prepared Output Format

Each prepare_*.py produces a sorted CSV:

timestamp,tag1[,tag2,...],field1[,field2,...]

• Sorted by device_id (tag combination), then timestamp
• Timestamp: epoch seconds (int64)
• Tags: string
• Fields: double

Plus a meta.json with schema and statistics.