next up previous
Next: 700hPa Geopotential Height Up: Reproducing the Grosswetterlagen Using Previous: Nearest Neighbour Classifiers

Surface Level Pressure

Figure 2 shows a graph of the proportion of correctly classified test patterns against values of k ranging from 1 to 100, for k-nearest neighbour classifiers used to reproduce the Grosswetterlagen catalogue on the basis of the first 26 principal components of the daily surface level pressure field. In each case a leave-one-out cross-validation strategy was employed to estimate the true generalisation error of the classifier. The optimal value for k was found to be 1, at which point 61.45% of the test patterns were classified correctly.


  
Figure 1: Graph of the proportion of correctly classified patterns against kfor k-nearest neighbour classifiers reproducing the Grosswetterlagen catalogue on the basis of the first 26 principal components of the daily surface level pressure field.
\begin{figure}
\begin{centering}
\includegraphics[scale=0.6]{results/parta/knn.eps}
\end{centering}\end{figure}

Table 1 shows a confusion matrix for the optimal 1-nearest neighbour classifier, again using a leave-one-out cross validation approach to estimate the true generalisation performance.


 
Table 1: Confusion matrix for a 1-nearest neighbour classifier reproducing the Grosswetterlagen catalogue on the basis of the first 26 principal components of the daily surface level pressure field.
  Model Classification
 
\begin{sideways}Wa \end{sideways}

\begin{sideways}Wz \end{sideways}

\begin{sideways}Ws \end{sideways}

\begin{sideways}Ww \end{sideways}

\begin{sideways}SWa \end{sideways}

\begin{sideways}SWz \end{sideways}

\begin{sideways}NWa \end{sideways}

\begin{sideways}NWz \end{sideways}

\begin{sideways}HM \end{sideways}

\begin{sideways}BM \end{sideways}

\begin{sideways}TM \end{sideways}

\begin{sideways}Na \end{sideways}

\begin{sideways}Nz \end{sideways}

\begin{sideways}HNa \end{sideways}

\begin{sideways}HNz \end{sideways}

\begin{sideways}HB \end{sideways}

\begin{sideways}TrM \end{sideways}

\begin{sideways}NEa \end{sideways}

\begin{sideways}NEz \end{sideways}

\begin{sideways}HFa \end{sideways}

\begin{sideways}HFz \end{sideways}

\begin{sideways}HNFa\end{sideways}

\begin{sideways}HNFz\end{sideways}

\begin{sideways}SEa \end{sideways}

\begin{sideways}SEz \end{sideways}

\begin{sideways}Sa \end{sideways}

\begin{sideways}Sz \end{sideways}

\begin{sideways}TB \end{sideways}

\begin{sideways}TrW \end{sideways}

\begin{sideways}U \end{sideways}
  Wa 372 72 1 4 8 5 13 8 31 47 0 0 3 2 4 5 7 2 3 2 0 0 0 2 0 4 0 2 7 2
  Wz 77 1075 20 11 28 27 11 38 24 65 3 1 15 3 10 2 31 1 3 3 1 0 1 1 4 1 3 9 46 10
  Ws 0 19 206 2 2 4 1 4 0 3 1 4 2 1 1 1 4 0 1 2 0 0 1 0 1 0 1 1 6 2
  Ww 5 13 2 133 5 3 1 2 4 8 0 1 2 0 0 0 2 1 1 5 3 0 0 1 3 3 0 1 8 1
  SWa 8 19 1 7 159 11 2 1 10 26 0 0 0 1 0 0 0 0 0 1 0 0 1 2 0 8 2 2 7 9
  SWz 3 31 9 4 12 212 1 1 6 9 1 0 2 2 1 1 1 1 1 2 2 0 2 3 0 2 1 6 4 8
  NWa 19 12 1 0 3 2 134 5 6 11 1 2 3 1 3 9 2 2 1 0 1 0 1 1 0 1 1 0 1 1
  NWz 17 47 3 3 2 0 10 227 8 11 2 2 17 2 5 9 15 1 1 7 0 0 0 1 0 0 1 0 11 10
  HM 32 20 1 3 13 8 9 5 279 66 0 1 1 4 2 15 4 1 1 15 2 2 2 7 1 10 4 2 8 5
  BM 42 56 3 7 15 9 18 16 56 634 8 5 9 3 1 18 16 5 9 10 7 5 4 13 4 9 3 2 20 8
  TM 2 8 2 1 0 0 1 4 0 9 144 0 5 1 4 2 14 3 1 1 3 1 4 4 3 0 0 6 11 3
  Na 0 2 4 1 0 0 1 2 1 4 1 33 1 2 2 0 1 3 0 1 1 0 2 0 0 0 0 0 3 0
  Nz 4 13 2 1 1 1 5 12 5 10 4 1 158 4 2 13 3 2 5 1 0 0 2 0 1 0 0 3 2 3
  HNa 1 1 3 0 1 2 2 0 2 3 2 1 6 108 2 6 1 1 1 3 1 5 2 2 0 0 1 1 1 3

\begin{sideways}\makebox[0pt]{True Classification}
\end{sideways}
HNz 3 6 2 0 1 0 3 2 5 4 3 1 1 3 110 2 3 3 1 2 2 3 5 2 0 0 0 2 2 5
  HB 5 2 0 0 0 3 12 5 14 28 4 3 8 9 7 186 6 6 3 7 2 5 1 2 0 1 0 1 2 5
  TrM 8 30 1 4 2 0 1 14 2 16 14 1 9 1 5 7 215 1 13 0 1 0 0 3 3 0 1 7 28 7
  NEa 1 2 0 1 1 0 1 2 0 7 3 1 0 0 0 5 1 88 4 4 5 3 1 1 0 0 0 0 1 1
  NEz 2 1 1 0 0 0 2 2 1 15 1 2 3 0 2 5 4 6 90 5 2 2 0 2 3 0 0 0 1 2
  HFa 3 4 3 3 0 2 0 1 6 16 2 1 1 3 0 5 2 14 3 197 9 5 6 9 6 2 2 1 1 2
  HFz 0 6 2 3 0 2 0 0 2 7 8 2 0 0 2 1 1 4 1 13 106 0 5 5 10 1 1 1 5 0
  HNFa 0 1 0 1 1 1 0 0 2 2 2 0 1 2 3 3 0 4 4 6 1 84 6 1 0 0 0 0 4 1
  HNFz 0 0 2 1 1 1 2 0 2 3 5 1 3 1 2 1 0 5 2 6 2 5 94 1 2 1 0 3 5 1
  SEa 2 1 0 1 0 1 2 0 9 6 1 0 0 2 3 2 2 0 0 7 3 0 1 113 5 10 2 3 10 4
  SEz 0 2 3 5 0 3 0 1 1 5 3 0 0 0 0 2 3 1 2 5 7 0 1 4 70 3 0 2 5 1
  Sa 3 1 0 2 9 2 0 0 9 11 1 0 0 1 1 0 2 0 0 1 1 3 0 8 3 96 2 3 2 3
  Sz 1 3 2 0 5 0 2 0 2 0 2 1 0 1 0 0 1 0 0 1 1 0 0 1 0 6 41 1 1 1
  TB 1 9 2 3 3 4 0 0 1 4 2 0 2 1 1 0 2 0 0 1 2 1 1 2 0 1 1 167 10 6
  TrW 7 30 7 6 10 9 2 9 6 20 12 1 4 0 3 1 23 1 3 2 3 4 5 7 2 4 0 12 287 7
  U 7 13 3 2 5 11 3 7 8 16 3 0 4 4 5 5 11 2 3 3 2 0 1 2 0 4 1 5 7 18
 


next up previous
Next: 700hPa Geopotential Height Up: Reproducing the Grosswetterlagen Using Previous: Nearest Neighbour Classifiers
Gavin Cawley ESE Faculty
1999-08-11