The only worse one is the one that throws out Rasmussen. (In science one learns that there are always outliers. If one wants an unbiased data set, you do not throw any of it out. The biases will cancel out, on average. Look up Gaussian distribution and standard deviation.)
All the othe electoral maps, none show Obama with a majority EV. The median one may be TPM, or the Princeton EV map.
However, none of them are meaningful at all until polling accelerates as we approach the election. The polls on which all of the EV projections are based are too old, too few, and way too speculative. The error bars are wide and one inevitably chooses the most optimistic as confirmation bias.
The best thing at this point in the game is to ignore the absolutes and look only at the changes.
Or! Make your own electoral map based only on all known and recent data or whatever criteria you choose. Just do not throw out any data, no matter how biased you perceive it to be. That's not allowed.