So, I ran my little predictive algorithm against the NCDC data, and I thought I’d throw these out so that I can either be publicly humiliated or take credit where due.
The following table predicts future temperature anomalies based on the analysis of how X-month trend line slopes change using the NCDC data. It uses weighting factors going back 11 years. Here are the results:
| 60-month | 120-month | 180-month | 240-month | 300-month | |
| May | 41.5 | 44.8 | 47.3 | 52.3 | 56.0 |
| June | 44.5 | 48.1 | 50.2 | 46.0 | 59.9 |
| July | 49.1 | 42.3 | 55.4 | 48.7 | 56.0 |
| August | 45.4 | 38.8 | 63.8 | 47.6 | 53.0 |
| September | 43.0 | 46.1 | 73.5 | 47.2 | 56.0 |
| October | 32.6 | 46.0 | 60.1 | 54.3 | 57.0 |
| November | 50.4 | 60.1 | 80.8 | 72.9 | 45.5 |
| December | 35.6 | 53.3 | 68.9 | 49.0 | 66.0 |
| January | 58.4 | 58.90 | 78.9 | 84.5 | 75.0 |
| February | 52.6 | 51.95 | 84.3 | 64.0 | 81.5 |
| March | 31.1 | 65.90 | 58.9 | 59.6 | 64.8 |
| April | 38.2 | 55.90 | 60.0 | 67.0 | 77.5 |
| 360-month | Average | Avg Excl Hi/Lo | Median | |
| May | 50.0 | 48.6 | 48.6 | 48.7 |
| June | 52.0 | 50.1 | 49.1 | 49.1 |
| July | 50.5 | 50.3 | 50.9 | 49.8 |
| August | 52.5 | 50.2 | 49.6 | 50.0 |
| September | 47.5 | 52.2 | 49.2 | 47.4 |
| October | 44.5 | 49.1 | 50.4 | 50.1 |
| November | 46.0 | 59.3 | 57.4 | 55.3 |
| December | 41.0 | 52.3 | 52.3 | 51.1 |
| January | 63.0 | 69.8 | 68.9 | 69.0 |
| February | 57.0 | 65.2 | 63.8 | 60.5 |
| March | 44.0 | 54.0 | 56.8 | 59.2 |
| April | 52.0 | 58.4 | 58.7 | 58.0 |
So there you go. I’ll continue to track the results going forward to see how I do, and how far out this model seems to have predictive capability – if any.
