Projection vs Reality and Analysis. A Detailed Analysis of my Projection for the 2015 Election Compared to the Actual Result, and What Happened?

Now that the election is over, and I have had an opportunity to decompress a little bit from my disgust and dismay at seeing yet another person named Trudeau elected to the PMO to ruin our country, it’s time to put on my analytical hat and see how well  my Patented Projection Prognosticator worked towards the election this time around.

First, the qualitative:

Projected outcome: Liberal Victory
Actual outcome: Liberal Victory

Okay, good so far.  Canada is worse-off for it, but my projection was accurate.

Projected Parliament: Liberal minority
Actual Parliament: Liberal majority

Okay, not so good there.  Majority vs Minority is a huge miss, but, to be fair, the pollsters were projecting the same kind of outcome as me, and since this is an algorithm, I guess it comes down to “garbage in = garbage out.”

Digging a bit deeper, recall how I offered a set of possible results:  Assured, Probable, Possible, and Long Shot.   Comparing those numbers we get this:

Comparison: Projected results vs Actual Results
(Expand the Items to view the tables)


National

Table 1: National

Party Assured Probable Possible Long Shot Max Projected Actual Relative Error
LPC 69 67 112 47 295 136 184 -9.20%
CPC 34 91 51 21 197 125 99 5.95%
NDP 6 64 46 29 145 70 44 6.81%
BQ 0 6 19 17 42 6 10 -1.15%
GP 0 1 0 0 1 1 1 0.00%
Other 0 0 1 0 1 0 0 0.00%


Atlantic

Quebec

Table 6: Quebec

Party Assured Probable Possible Long Shot Max Projected Actual Relative Error
LPC 6 22 23 25 76 28 40 -3.17%
CPC 1 14 5 11 31 15 12 0.86%
NDP 0 29 24 10 63 29 16 3.67%
BQ 0 6 19 17 42 6 10 -1.15%
GP 0 0 0 0 0 0 0 0.00%
Other 0 0 0 0 0 0 0 0.00%

Ontario

Table 7: Ontario

Party Assured Probable Projected Possible Long Shot Max Actual Projected Seat % Actual Seat % Relative Error
LPC 42 22 64 47 1 112 53 18.93% 15.68% 2.81%
CPC 5 35 40 18 3 61 21 11.83% 6.21% 5.29%
NDP 4 13 17 5 2 24 5 5.03% 1.48% 3.50%
GP 0 0 0 0 0 0 0 0.00% 0.00% 0.00%
Other 0 0 0 0 0 0 0 0.00% 0.00% 0.00%

Prairies

British Columbia

Table 11: British Columbia

Party Assured Probable Projected Possible Long Shot Max Actual Projected Seat % Actual Seat % Relative Error
LPC 3 7 10 18 10 38 17 2.96% 5.03% -1.97%
CPC 2 16 18 11 5 34 10 5.33% 2.96% 2.30%
NDP 2 11 13 6 8 27 14 3.85% 4.14% -0.28%
GP 0 1 1 0 0 1 1 0.30% 0.30% 0.00%
Other 0 0 0 0 0 0 0 0.00% 0.00% 0.00%

North

Assured: The leading candidate in the projected result is further ahead than any other candidate beyond the poll’s margin of error.
Probable: The leading candidate in the projected result is further ahead of at least one other candidate by less than the poll’s margin of error.
Possible: The candidate for a given party has a range of possible support where at least some of the range shows a possible win.
Long Shot: Same as “Possible” however the candidate is ranked behind at least one other candidate with a possible win.
Relative Error measures how far from the actual result the prediction was.  A negative number means the projection was low, and a positive number means it was high.  It is calculated using this formula, which eliminates division by zero errors: Error_{relative}=\frac{projected - actual}{338 + actual}\times 100\%

As you can see, the results were all within the range that I projected for each party.

This means that, while the Uniform Distribution Method of predicting seats (look it up), is a good estimation tool, it still can only be as good as the polling data available.

To get a better look at the accuracy of the predictor, let’s compare the popular vote in each region against the projected vote.  To do this, we will have to compare percentages rather than actual votes, because the number of votes cast between 2011 and 2015 is different.

Remember as well, the projection is based on aggregating four polls, all released on October 17:

Ekos, Forum, Mainstreet, Ipsos,  so the projected result for each riding below will be based on that aggregation, with the aggregated margin of error, calculated this way:

\sigma(region) _{total}= \sqrt{\sigma(region)_{Ekos}^2 + \sigma(region)_{Forum}^2 + \sigma(region)_{Mainstreet}^2 + \sigma(region)_{Ipsos}^2}

The aggregate support for each region is calculated like this.

Assume the Polls are identified like this:

1. Ekos
2. Forum
3. Mainstreet
4. Ipsos

In reality it doesn’t really matter what order the polls are in, just that they all get included…

Percent Support_{region,aggregate} = \frac{\sum\limits_{Poll = 1}^4(Percent Support_{(party,region) Poll} \times Sample Size_{Region,Poll})}{\sum\limits_{Poll = 1}^4 (SampleSize_{region,Poll})}

This gives us the aggregated percent support for each party in each region, from which we can then figure out the results for each individual constituency.

For the purposes of the Prediction Machine, when aggregating the results, I had to make sure that the results for each polling firm were consolidated into like regions, which was a quick and simple calculation.

So we end up with aggregated data for:

Atlantic
Quebec
Ontario
Manitoba/Saskatchewan
Alberta
British Columbia/Northwest Territories/Nunavut/Yukon Territory.

The territories really have no polling data to speak of, since their population is so sparse, so I just grouped them into the BC results. There wasn’t much impact either way.

So, with that in mind, here’s the regional results compared to the aggregated projection I posted on October 17th, for each party, along with the error in the seat projections.


National

Table 15: National Vote Comparison to Projected Vote

Party 2011% Projected Support 2015% Projection Error (%) 2011 Votes 2015 Votes Vote Change % Vote Change
LPC 18.90% 36.83% 39.47% -1.89% 2783117 6930136 4147019 149.01%
CPC 39.63% 31.74% 31.89% -0.12% 5835315 5600496 -234819 -4.02%
NDP 30.65% 21.49% 19.71% 1.49% 4512430 3461262 -1051168 -23.29%
BQ 6.05% 4.29% 4.66% -0.35% 891432 818652 -72780 -8.16%
GP 3.89% 4.33% 3.45% 0.85% 572113 605864 33751 5.90%
Other 0.88% 0.35% 0.81% -0.46% 129703 142943 13240 10.21%
Atlantic

Table 16: Atlantic Canada Vote Comparison to Projected Vote

Party 2011% Projected Support 2015% Projection Error (%) 2011 Votes 2015 Votes Vote Change % Vote Change
LPC 29.32% 52.60% 58.70% -3.84% 333175 768548 435373 130.67%
CPC 37.87% 20.89% 19.02% 1.58% 430349 248959 -181390 -42.15%
NDP 29.52% 19.43% 17.95% 1.25% 335451 235029 -100422 -29.94%
GP 2.99% 3.39% 3.54% -0.15% 33973 46367 12394 36.48%
Other 0.31% 0.31% 0.79% -0.48% 3520 10354 6834 194.15%
Quebec

Table 17: Quebec Vote Comparison to Projected Vote

Party 2011% Projected Support 2015% Projection Error (%) 2011 Votes 2015 Votes Vote Change % Vote Change
LPC 14.16% 29% 35.73% -4.69% 538458 1512676 974218 180.93%
CPC 16.52% 22% 16.73% 4.26% 627965 708225 80260 12.78%
NDP 42.90% 25% 25.35% 0.11% 1630877 1073212 -557665 -34.19%
BQ 23.45% 19% 19.34% -0.61% 891432 818652 -72780 -8.16%
GP 2.11% 4% 2.27% 1.66% 80402 96070 15668 19.49%
Other 0.86% 1% 0.59% -0.05% 32589 24858 -7731 -23.72%
Ontario

Table 18: Ontario Vote Comparison to Projected Vote

Party 2011% Projected Support 2015% Projection Error (%) 2011 Votes 2015 Votes Vote Change % Vote Change
LPC 25.32% 43% 44.76% -1.06% 1400313 2923791 1523478 108.80%
CPC 44.43% 33% 35.01% -1.84% 2457483 2287179 -170304 -6.93%
NDP 25.62% 19% 16.60% 2.27% 1417442 1084555 -332887 -23.49%
GP 3.75% 3% 2.90% 0.50% 207444 189373 -18071 -8.71%
Other 0.88% 0% 0.73% -0.33% 48845 47816 -1029 -2.11%
Manitoba/Saskatchewan

Table 19: MB/SK Vote Comparison to Projection

Party 2011% Projected Support 2015% Projection Error (%) 2011 Votes 2015 Votes Vote Change % Vote Change
LPC 12.70% 29% 34.66% -4.25% 120162 399589 279427 232.54%
CPC 54.86% 40% 42.68% -1.74% 519114 492116 -26998 -5.20%
NDP 28.94% 24% 19.22% 4.04% 273851 221570 -52281 -19.09%
GP 3.15% 5% 2.66% 2.47% 29784 30669 885 2.97%
Other 0.36% 0% 0.78% -0.54% 3410 8994 5584 163.75%
Alberta

Table 20: Alberta Vote Comparison to Projection

Party 2011% Projected Support 2015% Projection Error (%) 2011 Votes 2015 Votes Vote Change % Vote Change
LPC 9.26% 29% 24.55% 3.53% 129311 473661 344350 266.30%
CPC 66.82% 55% 59.54% -3.06% 932770 1148649 215879 23.14%
NDP 16.82% 14% 11.62% 2.24% 234731 224198 -10533 -4.49%
GP 5.23% 3% 2.54% 0.58% 73054 48908 -24146 -33.05%
Other 1.86% 0% 1.75% -1.50% 26021 33781 7760 29.82%
British Columbia/Territories

Table 21: BC/Territories Vote Comparison to Projection

Party 2011% Projected Support 2015% Projection Error (%) 2011 Votes 2015 Votes Vote Change % Vote Change
LPC 13.69% 32% 35.47% -2.42% 261698 851871 590173 225.52%
CPC 45.37% 32% 29.79% 1.83% 867634 715368 -152266 -17.55%
NDP 32.43% 26% 25.93% 0.09% 620078 622698 2620 0.42%
GP 7.71% 9% 8.10% 0.37% 147456 194477 47021 31.89%
Other 0.80% 0% 0.71% -0.58% 15318 17140 1822 11.89%

Each polling firm has weighted the data to ensure that the demographics are represented appropriately.  I have used the weighted data for this calculation, and, as well, I have included “leaners” with the data, when the polling firm has done so.

What we see in this table is the relative error in the projected support is less than 4.05% in all cases.  That’s pretty good – and means the projected support is fairly accurate compared to the aggregated polls.  So, what we need to do is look at the actual riding-by-riding results, and determine where the prediction machine went wrong.

Overall, the Prediction Machine was accurate in 265 of 338 constituencies.  That’s 78.4%.  It’s pretty good; but let’s look a bit closer, and examine where things went wrong.

Because we’re dealing with 338 constituencies, or Electoral Districts, as Elections Canada likes to call them, it’s probably easier to break things down into groups.  Above, I talked about Assured, ProbablePossible, and Long Shot.  Starting with Assured, which are seats the Prediction Machine suggested would be virtually guaranteed to be won by the projected party, we see this:

Out of 109 possible assured predictions, the Prediction Machine was accurate 106 times.  That’s 97.25% accuracy, which is an extremely good result.  But it’s also not that exciting – because an assured result means the predicted winner is further ahead than anyone else by more than the poll’s (or aggregated polls’) margin of error for the region.

So, where did the Prediction Machine get it wrong?

Markham-Unionville

Table 22: Inaccuracies in Assured Results: Marham - Unionville

Party 2011 Election Projected Result 2015 Election % Change Projection error
LPC 33.69% 51.60% 43.34% 7.22% 5.45%
CPC 45.83% 33.93% 49.37% 2.43% -11.53%
NDP 43.34% 10.16% 5.04% -9.86% 4.64%
GP 3.01% 2.67% 2.25% -0.74% 0.41%
Other 0.96% 0.48% 0% -0.95% 0.47%
Turnout 54.96% 60.38% 3.5%

What happened: The Conservative Party gained 6580 votes this election. This contrasts against a projected regional loss of 12%. The Liberal Party gained more votes than the Conservatives, however: 8347 votes. The NDP lost 3988 votes. The NDP and Liberal changes correspond roughly to the regional shifts shown in the aggregated poll, while the Conservatives seem to have gained mainly from a larger eligible voter base and higher turnout. Chances are this stems from the Conservative candidate having high name recognition and high popularity in the riding, neither of which are factors the Prediction Machine takes into account.

Nickel Belt

Table 23: Inaccuracies in Assured Results: Nickel Belt

Party 2011 Election Projected Result 2015 Election % Change Projection error
LPC 12.91% 31.97% 42.95% 26.61% -8.32%
CPC 25.73% 16.12% 16.80% -7.10 -0.58%
NDP 50.48% 48.62% 37.56% -8.59 7.44%
GP 2.56% 2.45% 2.49% -0.07% -0.03%
Other 0.12% -0.35% 0% -0.12% -0.35%
Turnout 61.93% 67.20% 3.25%

What happened: Turnout increased by 3.25%, or approximately 4800 votes. CPC Support changed roughly the same amount as the projection, so that follows. The NDP lost more support than expected, while the Liberals gained more support than expected. Once again, this shows the inherent volatility in the individual constituency projections. This election result in this riding is no surprise, given the area and the demographics.

Toronto-Danforth

Table 24: Inaccuracies in Assured Results: Toronto Danforth

Party 2011 Election Projected Result 2015 Election % Change Projection error
LPC 17.62% 35.53% 42.34% 21.02% -5.02%
CPC 14.32% 2.42% 9.86% -3.90% -7.26%
NDP 60.82% 54.43% 40.17% -12.83% 9.23%
GP 6.46% 6.12% 4.71% -1.64% 1.33%
Other 0.80% 0.32% 2.93% 2.11% -2.60%
Turnout 64.62% 72.04% 4.50%

What happened: Liberal support increased more than projected, while NDP did not increase as much as expected. CPC support also did not drop as much as expected. Likely, this comes from the nationwide surge in Liberal support that occurred in the last week of the campaign.

So, when it comes to the three ridings where the Prediction Machine missed an assured result, it came from one of two possible reasons – popularity of a specific candidate, or greater shifts in support that the polling firms didn’t pick up.  That isn’t necessarily the fault of the polling firms, because things were moving very quickly; but what that says to me is momentum is very important in elections, and in this particular election, the Liberal party, unfortunately, caught some momentum and it snowballed, picking up NDP supporters along the way.

Now, we can look at the probable results.  Probable, remember, is defined as the case where the second-placed candidate’s support is projected within the projected winner’s margin of error.   Out of 229 probable candidates, the Prediction Machine got it wrong 70 times, which is an accuracy of 69.43%.  That’s not too bad, but not great either.  Certainly, it’s lower than I’d like it to be.

ProvinceConstituencyLikely WinnerActual WinnerRelative Turnout Change (2011-2015) %BQ: ProjectedActualRPE*CPC: ProjectedActualRPE*GP: ProjectedActualRPE*LPC: ProjectedActualRPE*NDP: ProjectedActualRPE*Other: ProjectedActualRPE*
Newfoundland and LabradorSt. John's EastNDPLPC6.04%0.00%0.00%0.00%3.48%6.53%-2.87%1.53%1.10%0.42%31.06%46.77%-10.70%60.57%45.28%10.52%0.00%0.31%-0.31%
Nova ScotiaCentral NovaCPCLPC5.98%0.00%0.00%0.00%38.31%25.92%9.84%4.21%4.32%-0.10%37.24%58.25%-13.28%16.72%10.18%5.94%0.15%1.34%-1.18%
Nova ScotiaSackville--Preston--ChezzetcookNDPLPC7.90%0.00%0.00%0.00%13.21%14.88%-1.45%4.60%2.78%1.77%34.59%47.95%-9.03%44.21%34.39%7.30%0.01%0.00%0.01%
New BrunswickAcadie--BathurstNDPLPC4.31%0.00%0.00%0.00%-0.43%7.56%-7.43%0.47%2.33%-1.82%37.60%50.71%-8.70%58.99%39.40%14.05%0.00%0.00%0.00%
New BrunswickFundy RoyalCPCLPC6.36%0.00%0.00%0.00%40.97%37.07%2.84%5.12%3.89%1.19%33.72%40.87%-5.08%16.77%17.53%-0.64%0.02%0.63%-0.61%
New BrunswickNew Brunswick SouthwestCPCLPC5.85%0.00%0.00%0.00%39.69%38.56%0.81%5.50%4.95%0.52%36.85%43.92%-4.91%13.22%12.57%0.58%1.38%0.00%1.38%
New BrunswickTobique--MactaquacCPCLPC4.87%0.00%0.00%0.00%45.16%36.97%5.98%2.88%5.10%-2.11%39.52%46.63%-4.85%9.05%11.30%-2.03%0.02%0.00%0.02%
QuebecAbitibi--Baie-James--Nunavik--EeyouCPCNDP1.30%13.43%18.52%-4.30%27.74%9.30%16.87%5.74%2.26%3.41%25.66%32.09%-4.87%27.41%37.03%-7.02%-0.32%0.80%-1.12%
QuebecBerthier--MaskinongéLPCNDP2.63%26.17%25.72%0.36%17.90%10.18%7.01%4.19%1.55%2.59%27.27%20.31%5.79%23.83%42.24%-12.94%0.30%0.00%0.30%
QuebecThérèse-De BlainvilleNDPLPC3.55%23.01%26.93%-3.09%14.94%12.42%2.24%3.94%2.57%1.34%25.46%32.44%-5.27%32.63%24.90%6.19%-0.32%0.75%-1.07%
QuebecRimouski-Neigette--Témiscouata--Les BasquesBQNDP1.07%25.99%19.34%5.57%19.74%7.50%11.39%3.88%1.49%2.35%24.80%28.10%-2.57%25.57%42.96%-12.16%-0.32%0.61%-0.93%
QuebecChâteauguay--LacolleNDPLPC2.85%22.07%24.35%-1.83%17.21%11.19%5.41%3.76%1.99%1.73%25.40%39.05%-9.82%31.31%23.12%6.65%-0.09%0.29%-0.38%
QuebecChicoutimi--Le FjordCPCLPC1.17%23.98%20.54%2.85%30.75%16.41%12.32%3.37%2.06%1.28%20.95%31.13%-7.76%20.27%29.86%-7.39%0.35%0.00%0.35%
QuebecCompton--StansteadNDPLPC2.86%21.59%20.70%0.74%17.04%12.49%4.04%4.28%1.95%2.29%27.19%36.83%-7.04%29.87%27.42%1.92%-0.31%0.61%-0.92%
QuebecGatineauNDPLPC3.69%10.09%9.40%0.63%13.19%8.14%4.67%2.99%1.64%1.33%28.99%53.85%-16.16%44.72%26.55%14.36%-0.32%0.42%-0.74%
QuebecHull--AylmerNDPLPC3.78%3.84%6.51%-2.51%15.23%7.75%6.94%3.86%1.86%1.96%35.69%51.52%-10.45%41.37%31.36%7.62%-0.32%1.00%-1.31%
QuebecJolietteNDPBQ2.17%28.17%33.21%-3.78%14.88%10.00%4.43%5.84%2.43%3.32%21.59%28.18%-5.14%29.51%25.79%2.96%-0.32%0.38%-0.70%
QuebecJonquièreCPCNDP-6.28%14.42%23.30%-7.20%39.32%16.90%19.18%3.14%1.36%1.75%17.91%28.43%-8.20%25.09%29.20%-3.18%-0.21%0.80%-1.00%
QuebecLa Pointe-de-l'ÎleNDPBQ3.20%27.53%33.64%-4.58%12.83%7.98%4.49%3.73%2.04%1.66%25.29%28.57%-2.55%30.12%26.68%2.72%0.16%1.09%-0.92%
QuebecLa PrairieNDPLPC2.23%20.89%26.24%-4.24%14.63%11.91%2.43%3.30%2.15%1.13%28.42%36.47%-5.90%32.55%22.87%7.88%-0.14%0.35%-0.50%
QuebecLaurier--Sainte-MarieBQNDP-2.64%30.10%28.25%1.44%9.13%4.27%4.67%4.49%3.48%0.97%25.87%24.43%1.16%29.07%37.76%-6.31%1.00%1.81%-0.79%
QuebecManicouaganNDPBQ2.98%27.85%41.18%-9.45%16.81%10.24%5.96%4.36%1.64%2.68%20.70%29.41%-6.73%30.26%17.52%10.84%-0.32%0.00%-0.32%
QuebecMirabelNDPBQ3.38%26.18%31.37%-3.95%14.41%10.29%3.74%3.87%2.20%1.63%22.95%26.09%-2.49%32.13%30.05%1.60%0.12%0.00%0.12%
QuebecMontcalmNDPBQ1.89%26.60%36.58%-7.31%13.13%9.62%3.20%5.64%1.83%3.74%20.12%27.35%-5.68%34.49%23.45%8.94%-0.32%1.16%-1.47%
QuebecPontiacCPCLPC5.79%4.71%6.91%-2.06%31.37%13.93%15.31%3.48%1.74%1.72%30.38%54.54%-15.63%29.86%22.51%6.00%-0.15%0.38%-0.53%
QuebecPortneuf--Jacques-CartierNDPCPC3.58%15.64%10.74%4.43%5.18%43.95%-26.93%4.29%1.77%2.47%21.80%21.51%0.23%25.26%22.03%2.65%27.49%0.00%27.49%
QuebecQuébecNDPLPC1.82%23.12%18.82%3.62%22.95%22.17%0.64%4.03%2.90%1.10%24.21%28.79%-3.56%25.23%26.80%-1.24%0.11%0.52%-0.40%
QuebecRepentignyNDPBQ0.04%25.99%34.69%-6.46%12.64%10.81%1.64%3.58%1.90%1.64%23.12%27.27%-3.27%34.66%23.27%9.24%-0.32%2.04%-2.32%
QuebecRivière-des-Mille-ÎlesNDPLPC3.01%22.28%25.40%-2.48%15.48%10.50%4.51%4.22%1.95%2.23%26.29%32.42%-4.63%31.71%29.45%1.74%-0.32%0.29%-0.61%
QuebecRivière-du-NordNDPBQ2.86%23.36%32.04%-6.57%13.62%8.50%4.72%3.65%2.53%1.09%21.65%26.34%-3.72%37.71%30.13%5.82%-0.32%0.46%-0.78%
QuebecMarc-Aurèle-FortinNDPLPC3.56%19.43%21.43%-1.65%15.56%11.78%3.38%4.19%1.92%2.23%29.04%41.61%-8.87%31.57%23.26%6.74%-0.13%0.00%-0.13%
QuebecSaint-JeanNDPLPC4.37%25.66%24.81%0.68%15.85%10.85%4.51%4.37%2.12%2.20%24.04%33.15%-6.85%30.06%29.07%0.77%-0.32%0.00%-0.32%
QuebecSalaberry--SuroîtBQNDP2.59%28.91%28.23%0.53%17.87%9.95%7.20%3.68%1.42%2.23%23.33%29.23%-4.57%26.19%30.51%-3.31%-0.32%0.65%-0.97%
QuebecSheffordNDPLPC1.79%18.53%22.22%-3.02%19.83%12.78%6.26%3.74%2.37%1.34%24.19%38.96%-10.63%33.68%23.67%8.09%-0.32%0.00%-0.32%
QuebecSherbrookeBQNDP2.05%30.51%20.46%8.35%14.39%9.42%4.54%3.60%2.00%1.57%25.00%29.81%-3.71%26.06%37.33%-8.21%0.11%0.99%-0.87%
QuebecTerrebonneNDPBQ3.91%25.94%33.00%-5.31%13.25%11.31%1.74%4.54%1.79%2.70%22.12%27.99%-4.59%34.13%25.61%6.78%-0.32%0.30%-0.62%
OntarioBay of QuinteCPCLPC5.32%0.00%0.00%0.00%39.90%34.29%4.18%3.44%2.20%1.21%38.79%50.65%-7.87%16.64%12.21%3.95%0.05%0.64%-0.59%
OntarioBrantford--BrantLPCCPC3.71%0.00%0.00%0.00%36.51%40.85%-3.08%2.87%2.49%0.36%36.81%30.65%4.71%22.58%24.91%-1.87%0.05%1.09%-1.03%
OntarioBurlingtonCPCLPC4.17%0.00%0.00%0.00%42.24%42.52%-0.19%3.20%2.43%0.75%41.15%46.01%-3.33%12.48%9.04%3.15%-0.26%0.00%-0.26%
OntarioCambridgeCPCLPC3.54%0.00%0.00%0.00%40.84%38.66%1.58%3.26%3.21%0.05%33.35%43.21%-6.88%21.54%13.82%6.79%-0.19%1.11%-1.28%
OntarioDavenportNDPLPC6.31%0.00%0.00%0.00%2.31%10.55%-7.45%3.08%3.09%0.00%45.81%44.26%1.07%47.35%41.36%4.23%0.27%0.74%-0.47%
OntarioEssexCPCNDP4.92%0.00%0.00%0.00%36.36%35.68%0.50%2.09%1.89%0.20%32.17%20.88%9.34%28.53%41.43%-9.12%-0.34%0.13%-0.47%
OntarioHaldimand--NorfolkLPCCPC3.22%0.00%0.00%0.00%39.04%44.07%-3.50%2.97%3.32%-0.34%42.83%36.62%4.54%13.61%13.63%-0.02%0.38%2.35%-1.93%
OntarioHamilton East--Stoney CreekNDPLPC3.72%0.00%0.00%0.00%25.02%25.39%-0.29%2.51%2.57%-0.06%31.49%38.83%-5.29%37.88%32.72%3.89%1.91%0.49%1.41%
OntarioHastings--Lennox and AddingtonCPCLPC3.51%0.00%0.00%0.00%42.95%41.78%0.82%3.09%2.96%0.12%34.98%42.55%-5.31%17.62%12.71%4.36%0.18%0.00%0.18%
OntarioKitchener South--HespelerCPCLPC4.61%0.00%0.00%0.00%39.30%36.35%2.16%3.66%3.66%0.00%36.84%41.93%-3.59%19.39%15.72%3.18%-0.37%2.34%-2.64%
OntarioNewmarket--AuroraCPCLPC3.49%0.00%0.00%0.00%42.15%42.60%-0.32%4.04%2.36%1.65%41.61%45.18%-2.46%9.42%8.51%0.84%1.58%1.35%0.23%
OntarioNiagara CentreNDPLPC3.42%0.00%0.00%0.00%27.17%29.70%-1.95%2.14%2.42%-0.27%32.27%35.72%-2.54%36.80%31.44%4.08%0.43%0.72%-0.29%
OntarioNorthumberland--Peterborough SouthCPCLPC3.65%0.00%0.00%0.00%41.46%39.55%1.37%4.34%3.13%1.18%39.00%42.53%-2.47%14.21%14.79%-0.51%-0.21%0.00%-0.21%
OntarioOttawa CentreNDPLPC5.90%0.00%0.00%0.00%9.73%14.53%-4.19%4.70%2.98%1.67%38.03%42.63%-3.22%45.79%38.48%5.28%0.58%1.38%-0.80%
OntarioSt. CatharinesCPCLPC4.44%0.00%0.00%0.00%38.75%37.61%0.83%3.49%2.58%0.89%38.60%43.15%-3.18%17.57%16.51%0.92%0.40%0.15%0.25%
OntarioSpadina--Fort YorkNDPLPC5.74%0.00%0.00%0.00%8.96%15.72%-5.84%3.83%2.09%1.71%42.29%54.50%-7.90%43.27%27.40%12.45%0.46%0.29%0.17%
OntarioSudburyNDPLPC3.82%0.00%0.00%0.00%16.44%21.09%-3.83%2.65%3.04%-0.38%35.89%47.42%-7.82%43.55%27.79%12.33%0.27%0.66%-0.38%
OntarioThunder Bay--Rainy RiverNDPLPC4.88%0.00%0.00%0.00%15.31%21.11%-4.79%2.11%5.24%-2.97%39.61%44.06%-3.09%42.27%29.59%9.79%-0.48%0.00%-0.48%
OntarioThunder Bay--Superior NorthNDPLPC6.24%0.00%0.00%0.00%17.75%17.46%0.24%2.68%13.82%-9.79%34.38%44.85%-7.23%43.78%23.22%16.69%0.22%0.64%-0.42%
OntarioWhitbyCPCLPC4.47%0.00%0.00%0.00%46.91%42.10%3.38%4.64%2.17%2.42%32.40%44.88%-8.62%15.04%10.39%4.21%-0.17%0.46%-0.62%
ManitobaCharleswood--St. James--Assiniboia--HeadingleyCPCLPC6.03%0.00%0.00%0.00%42.90%39.17%2.68%5.97%2.90%2.98%34.63%51.90%-11.37%15.22%6.03%8.67%-0.13%0.00%-0.13%
ManitobaKildonan--St. PaulCPCLPC6.09%0.00%0.00%0.00%43.81%39.86%2.83%4.69%1.76%2.88%24.12%42.64%-12.99%25.14%14.30%9.49%0.82%1.43%-0.60%
ManitobaWinnipeg CentreNDPLPC8.91%0.00%0.00%0.00%12.98%12.36%0.55%9.10%4.07%4.83%27.30%54.51%-17.61%48.76%28.00%16.21%0.46%1.06%-0.60%
AlbertaCalgary CentreCPCLPC11.70%0.00%0.00%0.00%43.21%45.18%-1.36%8.23%2.20%5.90%38.81%46.65%-5.35%12.47%5.57%6.54%-1.62%0.40%-2.02%
AlbertaEdmonton Mill WoodsCPCLPC9.76%0.00%0.00%0.00%46.78%41.06%4.05%0.89%2.23%-1.30%31.42%41.23%-6.95%22.70%12.78%8.80%-0.69%2.70%-3.31%
British ColumbiaCloverdale--Langley CityCPCLPC7.85%0.00%0.00%0.00%44.17%34.79%6.96%5.41%4.15%1.21%30.31%45.48%-10.43%18.30%15.58%2.35%0.87%0.00%0.87%
British ColumbiaCoquitlam--Port CoquitlamCPCLPC7.70%0.00%0.00%0.00%42.32%31.94%7.87%5.12%3.86%1.21%26.78%35.09%-6.16%24.59%27.31%-2.14%0.23%1.79%-1.53%
British ColumbiaFleetwood--Port KellsCPCLPC0.36%0.00%0.00%0.00%34.62%29.29%4.12%3.36%2.37%0.96%34.18%46.87%-8.64%26.68%21.47%4.29%0.21%0.00%0.21%
British ColumbiaKelowna--Lake CountryCPCLPC8.95%0.00%0.00%0.00%45.17%39.76%3.87%9.57%0.00%9.57%29.98%46.15%-11.06%14.98%14.09%0.78%-0.67%0.00%-0.67%
British ColumbiaKootenay--ColumbiaCPCNDP6.39%0.00%0.00%0.00%36.88%36.78%0.07%7.22%6.51%0.66%22.00%19.48%2.10%32.46%37.23%-3.47%0.48%0.00%0.48%
British ColumbiaMission--Matsqui--Fraser CanyonCPCLPC0.93%0.00%0.00%0.00%41.27%34.88%4.74%5.43%5.13%0.28%26.92%37.25%-7.53%25.63%20.56%4.20%-0.22%2.17%-2.34%
British ColumbiaPitt Meadows--Maple RidgeCPCLPC8.12%0.00%0.00%0.00%41.63%31.40%7.78%5.68%4.22%1.39%23.89%33.76%-7.38%28.51%29.63%-0.86%-0.67%0.99%-1.65%
British ColumbiaSteveston--Richmond EastCPCLPC6.09%0.00%0.00%0.00%40.70%38.47%1.61%4.93%3.67%1.21%37.11%45.08%-5.49%16.44%12.14%3.84%-0.15%0.63%-0.78%
Northwest TerritoriesNorthwest TerritoriesNDPLPC6.99%0.00%0.00%0.00%18.89%18.01%0.75%3.85%2.82%1.00%36.95%48.34%-7.68%39.46%30.83%6.60%-0.11%0.00%-0.11%
* Relative Projection Error

Looking at the results where the Prediction Machine missed on the Probable winner, we see that, just like with the Assured results, turnout was, generally, higher than 2011, the Liberal support surged at the last minute at the cost of both the Conservatives and the NDP.

For completeness, I’ll include two more tables:  One for ridings where not the Possible candidate didn’t win, and another where not even the Long Shot contender won.  In that last circumstance, it represents where the Prediction Machine was completely wrong.

Table 26: Inaccuracies in Possible Winners

ProvinceConstituency NamePossible WinnerActual WinnerTurnout: Relative Change (%)BQ: ProjectedACTUALRPE*CPC: ProjectedActualRPE*GP: ProjectedActualRPE*LPC: ProjectedActualRPE*NDP: ProjectedActualRPE*OTHER: ProjectedActualRPE*
QuebecBerthier--MaskinongéBQNDP2.63%26.17%25.72%0.36%17.90%10.18%7.01%4.19%1.55%2.59%27.27%20.31%5.79%23.83%42.24%-12.94%0.30%0.00%0.30%
QuebecChicoutimi--Le FjordBQLPC1.17%23.98%20.54%2.85%30.75%16.41%12.32%3.37%2.06%1.28%20.95%31.13%-7.76%20.27%29.86%-7.39%0.35%0.00%0.35%
QuebecPortneuf--Jacques-CartierLPCCPC3.58%15.64%10.74%4.43%5.18%43.95%-26.93%4.29%1.77%2.47%21.80%21.51%0.23%25.26%22.03%2.65%27.49%0.00%27.49%
QuebecSaint-JeanBQLPC4.37%25.66%24.81%0.68%15.85%10.85%4.51%4.37%2.12%2.20%24.04%33.15%-6.85%30.06%29.07%0.77%-0.32%0.00%-0.32%
OntarioEssexLPCNDP4.92%0.00%0.00%0.00%36.36%35.68%0.50%2.09%1.89%0.20%32.17%20.88%9.34%28.53%41.43%-9.12%-0.34%0.13%-0.47%
ManitobaKildonan--St. PaulNDPLPC6.09%0.00%0.00%0.00%43.81%39.86%2.83%4.69%1.76%2.88%24.12%42.64%-12.99%25.14%14.30%9.49%0.82%1.43%-0.60%
British ColumbiaPitt Meadows--Maple RidgeNDPLPC8.12%0.00%0.00%0.00%41.63%31.40%7.78%5.68%4.22%1.39%23.89%33.76%-7.38%28.51%29.63%-0.86%-0.67%0.99%-1.65%
*Relative Prediction Error
ProvinceConstituency NameLong ShotActual WinnerTurnout: Relative Change (%)BQ: ProjectedACTUALRPE*CPC: ProjectedActualRPE*GP: ProjectedActualRPE*LPC: ProjectedActualRPE*NDP: ProjectedActualRPE*OTHER: ProjectedActualRPE*
QuebecPortneuf--Jacques-CartierBQCPC3.58%15.64%10.74%4.43%5.18%43.95%-26.93%4.29%1.77%2.47%21.80%21.51%0.23%25.26%22.03%2.65%27.49%0.00%27.49%

So, what does it all mean?

Well, the bottom line is this – first, the Prediction Machine, I think, is probably as accurate as it can be.  There may be a few tweaks to the prediction formula; and I’ll experiment with them in time.  But overall, we are dealing with a program that tries to model an election from a less-than-perfectly-accurate survey, or aggregate of surveys, of the population.

It’s a prediction based on a prediction.

80% accuracy, I have to say, is pretty good, considering the variables at play here, and, I think, in the majority of election campaigns, I think I would have probably called the result pretty accurately.  Most predictions were for a Liberal minority as well, so, while I was wrong in calling the size of the Parliament, I think the actual results I put forth were quite accurate.

So, I think I can safely say that my Prediction Machine will call the election with about 80% accuracy; meaning that 270 out of the 338 seats will be predicted correctly.  That’s means that, out of my projected results, there are 2,678,521,876,251,498,576,491,365,815,949,827,268,919,378,444,242,535,468,031,460,203,168,021,780 possible ways 270 seats are accurately projected.  That may seem like a ridiculously big number, but consider we’re dealing with cominatorial mathematics here.

It doesn’t really matter, however , how many possible outcomes there are with 270 accurately-predicted seats, only the one that the Prediction Machine kicks out is important; compared to the actual result from election day.  In that context — and we’ll have to wait up to four years to see — I expect that I’ll be calling the election result within margin of error and with 80% accuracy again.

 

About Steven Britton

Steve is a freelance programmer, partial billionaire, dad, Recovering Atheist, Conservative, and occasionally prolific blogger.