As we enter the week of the election, uncertainty still reigns supreme. No-one knows with any certainty who will form the next government, or even whether we'll end up having a second election this year.
Luckily for us, uncertainty is built into the way we have modelled the General Election, so this plays into our hands more than it does those predicting the election using traditional methods.
Whilst the changes for Alliance and Ulster Unionists look dramatic, a look at the absolute values show that these are minor changes exaggerated due to these parties' low counts.
Therefore, the most dramatic change really occurs with UKIP, and we see that they are expected to win one less seat than they were a week ago (it should be noted that both predictions are still much higher than the 3-4 predicted elsewhere).
Elsewhere, the chances are relatively glacial, although the Conservatives will be pleased to see their slender lead over Labour stretch by an extra seat, making them strong favourites to earn the first chance of forming a coalition. The SNP and the Liberal Democrats continue their respective upwards and downwards trajectories to no great surprise.
So, which seats have contributed the changes we have seen?
To compare changes in constituencies, we will look at the swing - this we will measure as the highest percentage change for any party between our current values and those last week.
There are eight constituencies where the swing is over 10%, and these are listed below. The colour in the final column indicates towards which party the swing is to.
Most significant for UKIP is the change in Thanet South, where Nigel Farage is standing. Throughout the campaign this has been a very likely UKIP gain in our model, but the latest swing sees this become a much closer marginal seat. On a lesser scale, Labour's increase in Peterborough sees that seat become closer. UKIP's decreases elsewhere see their challenger status diminish, which is also the status of the Liberal Democrats in Watford.
However, the two seats which see the biggest change are the most interesting, as these are the only seats where the party in the lead has changed. The first of these is Dumfriesshire, Clydesdale & Tweeddale, where the SNP continue their rapid rise by becoming unlikely favourites in the one Conservative-held constituency, following a week where polls have predicted they could win every single Scottish seat (although within our model, as with our Scotland-specific update, they are still not very likely to make much headway in Orkney & Shetlands).
The other seat is Croydon Central, which has seen the lead switch from Labour to the Conservatives. This is due to an Ashcroft poll being published in the last week suggesting a slight change in voting intentions.
Unsurprisingly, as we updating our model at more frequent time points, we see that the changes between iterations are smaller. However, because of these smaller changes, we can also look at individual seats to see what is driving the changes in party totals.
We have seen a few seats which are clearly going to be marginals where the result is on a knife-edge, which will be of particular interest on election night. Next time we will be looking to see which are the other marginals.