added features. yes its late, dont judge me

toni/octave/functions.m

@@ -167,3 +167,80 @@ windowedData = {};
   end  
 end
 
+function features = featureCalculation(data)
+
+features = [];
+
+  for k = 1:numel(data)
+    for j = 1:numel(data{k}.samples)
+      for i = 1:numel(data{k}.samples{j}.raw)
+        for m = 1:numel(data{k}.samples{j}.raw{i}.wins)
+          currentWindow = data{k}.samples{j}.raw{i}.wins{m};
+          currentWindow = currentWindow(:,1);
+          
+          #autocorrelation on window. split into 5 evenly spaced bins (frequencies are evenly spaced, not number of values ;) ) and calculate mean of bin. 
+          [autoCorr] = xcorr(currentWindow);
+          [binNum, binCenter] = hist(autoCorr, 5); #define 5 bins for the data.
+          binSize = abs(binCenter(end-1) - binCenter(end));
+          binEdges = linspace(binCenter(1)-(binSize/2), binCenter(end)+(binSize/2), 6);
+          [binNumc, binIdx] = histc(autoCorr, binEdges);
+          binMeans = getBinMean(autoCorr, binIdx, 5);
+          
+          #calculate the root-mean-square (RMS) of the signal
+          rms = sqrt(mean(currentWindow.^2));
+          
+          #power bands 0.5 to 25hz (useful if the windows are greater then 4s and window sizes to 256, 512..) 
+          [powerBand, w] = periodogram(currentWindow); #fills up fft with zeros            
+          powerEdges = logspace(log10(0.5), log10(25), 10 + 2);  #logarithmic bin spaces for 10 bins
+          triFilter = getTriangularFunction(powerEdges, length(powerBand)*2 - 2);    
+          for l = 1:numel(triFilter)
+            filteredBand = triFilter{l} .* powerBand;
+            psd(l) = sum(filteredBand); #sum freq (no log and no dct)
+          end
+          
+          #put everything together
+          classLabel = k; #what class?
+          features = [features; classLabel, binMeans, rms, psd];
+        end
+      end 
+    end
+  end
+end
+
+function value = getBinMean(data, idx, numBins)
+
+
+value = [];
+  for i = 1:numBins
+    flagBinMembers = (idx == i);
+    binMembers = data(flagBinMembers);
+    
+   # if length(binMembers) == 0
+     # idx
+      #data
+     # input = 'balala'
+    #endif
+    
+    value(i) = mean(binMembers);
+  end  
+end
+
+#triangular functions. (edges of the triangles; num fft values -> nfft.)
+function triFilter = getTriangularFunction(edges, nfft)
+
+  #get idx of the edges within the samples. thanks to fft each sample represents a frequency. 
+  # idx * samplerate / nfft = hertz of that idx
+  for i = 1:length(edges)
+    edgesByIdx(i) = floor((nfft + 1) * edges(i)/100); #100hz is the samplerate
+  end
+  
+  #generate the triangle filters
+  triFilter = {};
+  for i = 1:length(edgesByIdx)-2
+    diffCnt = edgesByIdx(i+2) - edgesByIdx(i) + 1;
+    tmp = zeros(nfft/2 + 1, 1);
+    triVec = triang(diffCnt);
+    tmp(edgesByIdx(i):edgesByIdx(i+2)) = triVec;
+    triFilter{i} = tmp;
+  end
+end

toni/octave/rawPlot.m

@@ -54,7 +54,12 @@ magnitude is calculated using sqrt(x^2 + y^2 + z^2) for each sensor.
 windowedClasses = windowData(filteredClasses);
 
 #calculated features for the 5 signales (x, y, z, MG, PCA) of a sensor
-windowedFeatures = featureCalculation(windowedClasses);
+#{
+data structure of features
+label | feature #1 | 2 | 3 ... 
+
+#}
+features = featureCalculation(windowedClasses);
 
 #train svm
 

toni/octave/todo.txt

@@ -18,11 +18,11 @@ pca (erste pca) (projektion der x,y,z koordinaten auf den ersten eigenvektor)
 
 2) generiere features für die jeweiligen fenster aus den signalen
 autocorrelation features: 
-autocorrelation bins
+autocorrelation bins. jedes signal window wird in 5 frequenz bins unterteilt und von diesen der mean berechnet. wäre es nicht besser, für einen sensor feste bins für alle klassen zu haben? braucht aber wohl mehr bins, sonst ist der bereich zu klein?
 
 energy features:
 RMS
-power spectrum bin mangnitudes
+power spectrum (|fft|^2). features über logarithmisch überlappende vierrecksfilter, da dreieicksfilter hier keinen sinn ergeben, oder? Wie bei Audio noch eine DCT auf die Ergebnisse des Dreiecksfilters um unwichtige rauszuwerfen?
 
 statistical features:
 mean
@@ -34,3 +34,31 @@ interquartile range
 
 classification:
 SVM
+
+
+
+Abschluss-Dokument: 
+
+- was geht
+- was hat nicht funktioniert?
+
+Training:
+one leave out.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+