MissNet: Mining of Switching Sparse Networks for Missing Value Imputation in Multivariate Time Series

Implementation of MissNet, Kohei Obata, Koki Kawabata, Yasuko Matsubara, Yasushi Sakurai. The 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, KDD'24.

Requirements

We run all the experiments in python 3.8, see requirements.txt for the list of pip dependencies.

To install packages

pip install -r requirements.txt

or

pip install numpy
pip install pandas
pip install matplotlib
pip install seaborn

Datasets preparation

Synthetic datasets

Generate synthetic datasets with the code below.

cd  data
python generate.py

---

MotionCapture datasets

MotionCapture datasets are stored in dynammo.zip.

The original data can be downloaded from link. We changed the filename for convenience.

This is the original website link.

---

Motes dataset

Motes dataset is stored in motes.zip.

The original data can be downloaded from link.

Missing block generation

We generate missing blocks for the experiments.

cd  data
python conversion.py

Change missing_rate_test in conversion.py for full experiments (if necessary).

The data containing missing blocks will be stored in ./data/experiment.

Experiments

After the preparation of datasets and missing blocks, run below.

python Experiment.py --datasets dynammo
python Experiment.py --datasets motes
python Experiment.py --datasets synthetic/pattern

Citation

If you use this code for your research, please consider citing our KDD paper.