KNN Resampler

This notebook shows how KNN Resampler is used to create synthetic data. There are three methods available in synloc package. This notebook demonstrates two methods: LocalCov and LocalGaussianCopula. The first one draw synthetic values from multivariate normal distribution and the second one draws from the gaussian copula.

from synloc import LocalCov, LocalGaussianCopula, sample_trivariate_xyz

Data

df = sample_trivariate_xyz(1000)
df.head()

	x	y	z
0	0.001029	1.443241e-02	1.030596
1	0.000010	7.651150e-08	-0.402560
2	0.002199	8.689394e-01	9.819810
3	0.999379	1.780679e-01	1.473825
4	0.064769	9.160882e-01	9.113435

Using Multivariate Normal Distribution

We use LocalCov method to create synthetic data. The method uses k-nearest neighbors to create subsamples from nearest neighbors. Then, it estimates the covariance matrix of each subsample and draw synthetic values from the multivariate normal distribution.

syn = LocalCov(df, K = 20) # K is the subsample size.
df_syn = syn.fit()

100%|██████████| 1000/1000 [00:01<00:00, 684.34it/s]

After the synthesis complete, you can use the class method comparePlots to visualize the synthetic data and the original data.

syn.comparePlots(['x', 'y', 'z'])

Using Gaussian Copula

syn_g = LocalGaussianCopula(df, K = 20)
df_syn = syn_g.fit()

100%|██████████| 1000/1000 [00:05<00:00, 173.02it/s]

syn_g.comparePlots(['x', 'y', 'z'])