One of the main challenges of modern cosmology is to investigate the nature of dark energy in our Universe .
The properties of such a component are normally summarised as a perfect fluid with a ( potentially ) time-dependent equation-of-state parameter w ( z ) .
We investigate the evolution of this parameter with redshift by performing a Bayesian analysis of current cosmological observations .
We model the temporal evolution as piecewise linear in redshift between ‘ nodes ’ , whose w -values and redshifts are allowed to vary .
The optimal number of nodes is chosen by the Bayesian evidence .
In this way , we can both determine the complexity supported by current data and locate any features present in w ( z ) .
We compare this node-based reconstruction with some previously well-studied parameterisations : the Chevallier-Polarski-Linder ( CPL ) , the Jassal-Bagla-Padmanabhan ( JBP ) and the Felice-Nesseris-Tsujikawa ( FNT ) .
By comparing the Bayesian evidence for all of these models we find an indication towards possible time-dependence in the dark energy equation-of-state .
It is also worth noting that the CPL and JBP models are strongly disfavoured , whilst the FNT is just significantly disfavoured , when compared to a simple cosmological constant w = -1 .
We find that our node-based reconstruction model is slightly disfavoured with respect to the \Lambda CDM model .