TiGraMITe

"Tigramite is a time series analysis python module for linear and information-theoretic causal inference. With flexibly adaptable scripts it allows to reconstruct graphical models (conditional independence graphs) from discrete or continuously-valued time series based on a causal discovery algorithm, quantify interaction strengths with different measures, and create high-quality plots of the results.

The features:

• Analysis can be performed on multivariate time series. Further scripts allow sliding window or ensemble analyses
  
• Functions for custom preprocessing like anomalization, high/lowpass filters, masking of samples (e.g. winter months only), time-binning, ordinal pattern analysis, and more
  
• Different (conditional) measures of association (partial correlation, standardized regression, and conditional mutual information with different estimators)
  
• Fast computation through use of Cython; also fully parallelized script (mpi4py package necessary) available
  
• Significance testing via analytical tests or a shuffle test for conditional mutual information
  
• Flexible plotting scripts for publication quality presentation of results
  

https://github.com/jakobrunge/tigramite 
 
Detecting causal associations in large nonlinear time series datasets - https://arxiv.org/abs/1702.07007