Writing up results

This commit is contained in:
2017-08-23 09:44:00 +01:00
parent c030646d81
commit 613047f88d
+82 -43
View File
@@ -1476,7 +1476,7 @@ The final optimised model was generated using a total of 43 selected features. P
evaluations, resulting in 50 iterations using 20 particles. Final parameters
and selected features
for the chosen algorithms are detailed in table~\ref{OpParam}.\\
The final scores produced for this model, evaluated using the full dataset can
The final scores produced for this model, evaluated using the full dataset, can
be found in Table~\ref{TestSet} (Hidden test set scores), Table~\ref{LOGO}
(Leave-one-out scores) and
Table~\ref{KFCV} (Stratified cross-validation scores).
@@ -1496,6 +1496,8 @@ $Acc$ & $Se$ & $Sp$ \\ \midrule
\doublespacing
\caption{Leave-one-out scores}
\label{LOGO}
\footnotesize
All scores are an average of 10 iterations
\scriptsize
\centering
\begin{tabulary}{\linewidth}{LCCCCCCC}
@@ -1509,6 +1511,8 @@ $Sp$ & $0.3509\pm0.0264$ & $0.1127\pm0.012$ & $0.4571\pm0.0571$ & $0.2481\pm0
\begin{table}[H]
\caption{10-fold cross-validation score}
\footnotesize
All scores are an average of 10 iterations
\doublespacing
\label{KFCV}
\scriptsize
@@ -1599,11 +1603,31 @@ C: 4.2507 & C: 4.9452 & & C: 14.3611
\end{multicols}
\end{table}
Weighted specificity and weighted Accuracy measures
Leave-one-out cross-validation results are compareable to those of the highest
scoring algorithms in the challenge, however they are still low scores.~\parencite{Homsi2017, Bobillo2016}
10-fold cross-validation scores are at worst, around 12\% less than those of
the highest scoring models.~\parencite{Zabihi2016}
Hidden test set is the only score based on predictions where no samples had
previously been seen by the algorithm during optimisation. A similar score to
that off 10-fold cross validation suggests that chosen features and
hyperparameters generalise well to unseen data. If scores in cross validation
had been significantly higher than that of the hidden test set, it would
suggest that the model is tuning parameters and features in a way that only
benefits the score of the training set.
higher scores in 10-fold cross validation than those of Leave-one-out
cross-validation suggests that the algorithm is highly susceptible to degraded
results as a consequence of signal qualities varying from those of the training
set.
Computational cost was not considered, unlike other entries to the physionet
challenge
Could be used as cloud based system
Discussion on reasons for final selection of models
Discussion on final selection of models
Features were selected for their individual relevance to classification
problem, Naive Bayes treats features individually. Could explain why it
performed well
@@ -1612,6 +1636,10 @@ captured by SVMs
Discuss issues with database e
Due to the standard approach taken for scoring entries to the physionet
challenge, mimicked in this project, it was possible to directly compare results
to those of entries to the challenge. This aims to provide an understanding of
\section{Further Work}\label{FurtherWork}
Further research to be done into resampling - inclusion as hyperparameter in
optimization
@@ -1619,6 +1647,10 @@ Handle silent sections of audio such as those highlighted by Goda et.\
al~\parencite{Goda2016}
Synthesis of synthetic PCG signals
Particle swarm Would ideally be placed inside feature selection
MFCC features were overly symplistic, using coefficients diectly. Further
analysis of these features as demonstrated in other
algorithms~\parencite{Zabihi2016} may improve there impact on results.
% TODO: Consider talking about resampling using Homsi2016 method
\appendix
@@ -1629,50 +1661,57 @@ Particle swarm Would ideally be placed inside feature selection
\subsection{Table of Features}\label{appendixA}
\begin{table}[H]
\centering
\caption{My caption}
\doublespacing
\label{my-label}
\caption{Description of features}
\scriptsize
Feature sources include:~\parencite{Homsi2016, Schmidt2015, Liang1998,
Lerch2012}\\
`*' --- denotes feature is applied to S1, systolic, S2 and diastolic segments
respectively.
\onehalfspacing
\tiny
\label{my-label}
\begin{tabulary}{\linewidth}{LLLLL}
\toprule
Tag & Feature Name & Description & Ref. \\ \midrule
heartRate & Heart Rate & The number of beats per minute (BPM) & \\
m\_RR & Mean RR Interval & Average length of heart cycles & \\
sd\_RR & Std-dev RR Interval & Standard deviation of heart cycles & \\
mean\_Int* & Mean Segment Interval & Average length of segment & \\
sd\_Int* & Std-dev Segment Interval & Standard deviation of segment & \\
R\_SysRR & Systolic-RR Ratio & Ratio of Systolic interval to RR interval & \\
R\_DiaRR & Diastolic-RR Ratio & Ratio of Diastolic interval to RR interval & \\
R\_SysDia & Ratio of Systolic-RR/Diastolic-RR Ratios & Ratio of above ratios & \\
*ZeroX & Zero-crossing & Zero-crossing rate of a segment & \\
*RMS & Root Mean Square & The Root Mean Square of a segment & \\
*ShanEngy & Shannon Energy & The Averaged Shannon Energy Envelope of a segment & \\
*Dur & Duration & The duration of a segment & \\
*Max & Max & The peak value of an absolute segment & \\
*Mean & Mean & The mean value of a segment & \\
*Skew & Skewness & The temporal skewness of a segment & \\
*Kurt & Kurtosis & The temporal kurosis of a segment & \\
*Var & Variance & The variance of a segment & \\
*SampEnt & Sample Entropy & The sample entropy of a segment & \\
*ShanEnt & Shannon Entropy & The Shannon Entropy of a segment & \\
TPT* & Total Power (time) & The total power of a segment in the time domain & \\
TPF* & Total Power (frequency) & The total power of a segment in the frequency domain & \\
*Flat & Spectral Flatness & The flatness of a segment's frequency spectrum & \\
*Cent & Spectral Centroid & The centroid of a segment's frequency spectrum & \\
*Spread & Spectral Spread & The spread of a segment's frequency spectrum & \\
*MFCC\{$n$\} & Mel-frequency Cepstrum Coefficients & MFCC coefficient number $n$, where $n = \{1, \ldots, 13\}$ & \\
m\_Ratio\_SysRR & Mean RR/Systole interval ratio & Mean value of the interval ratios between systole and RR in each heart beat & \\
sd\_Ratio\_SysRR & Std-dev RR/Systole interval ratio & Std-dev value of the interval ratios between systole and RR in each heart beat & \\
m\_Ratio\_DiaRR & Mean RR/Diastole interval ratio & Mean value of the interval ratios between diastole and RR in each heart beat & \\
sd\_Ratio\_DiaRR & Std-dev RR/Diastole interval ratio & Std-dev value of the interval ratios between diastole and RR in each heart beat & \\
m\_Ratio\_SysDia & Mean Systole/Diastole interval ratio & Mean value of the interval ratios between systole and diastole in each heart beat & \\
sd\_Ratio\_SysDia & Std-dev Systole/Diastole interval ratio & Std-dev value of the interval ratios between systole and diastole in each heart beat & \\
D[1-5]Shan & Detail coefficient shannon entropy & Shannon entropy of DWT detail coefficient 1-5 & \\
A5Shan & Approximation coefficient shannon entropy & Shannon entropy of DWT approximation coefficient 1-5 & \\
\mbox{TotD[1-5]*Shan} & Total detail coefficient shannon entropy & Total Shannon entropy of DWT detail coefficient 1-5 across signal & \\
TotA5*Shan & Total approximation coefficient shannon entropy & Total Shannon entropy of DWT approximation coefficient 1-5 across signal & \\ \bottomrule
\hline
Tag & Feature Name & Description \\ \hline
heartRate & Heart Rate & The number of beats per minute (BPM) \\
m\_RR & Mean RR Interval & Average length of heart cycles \\
sd\_RR & Std-dev RR Interval & Standard deviation of heart cycles \\
mean\_Int* & Mean Segment Interval & Average length of segment \\
sd\_Int* & Std-dev Segment Interval & Standard deviation of segment \\
R\_SysRR & Systolic-RR Ratio & Ratio of Systolic interval to RR interval \\
R\_DiaRR & Diastolic-RR Ratio & Ratio of Diastolic interval to RR interval \\
R\_SysDia & Ratio of Systolic-RR/Diastolic-RR Ratios & Ratio of above ratios \\
*ZeroX & Zero-crossing & Zero-crossing rate of a segment \\
*RMS & Root Mean Square & The Root Mean Square of a segment \\
*ShanEngy & Shannon Energy & The Averaged Shannon Energy Envelope of a segment \\
*Dur & Duration & The duration of a segment \\
*Max & Max & The peak value of an absolute segment \\
*Mean & Mean & The mean value of a segment \\
*Skew & Skewness & The temporal skewness of a segment \\
*Kurt & Kurtosis & The temporal kurosis of a segment \\
*Var & Variance & The variance of a segment \\
*SampEnt & Sample Entropy & The sample entropy of a segment \\
*ShanEnt & Shannon Entropy & The Shannon Entropy of a segment \\
TPT* & Total Power (time) & The total power of a segment in the time domain \\
TPF* & Total Power (frequency) & The total power of a segment in the frequency domain \\
*Flat & Spectral Flatness & The flatness of a segment's frequency spectrum \\
*Cent & Spectral Centroid & The centroid of a segment's frequency spectrum \\
*Spread & Spectral Spread & The spread of a segment's frequency spectrum \\
*MFCC\textbackslash\{$n$\} & Mel-frequency Cepstrum Coefficients & MFCC coefficient number $n$, where $n = \{1, \ldots, 13\}$ \\
m\_Ratio\_SysRR & Mean RR/Systole interval ratio & Mean value of the interval ratios between systole and RR in each heart beat \\
sd\_Ratio\_SysRR & Std-dev RR/Systole interval ratio & Std-dev value of the interval ratios between systole and RR in each heart beat \\
m\_Ratio\_DiaRR & Mean RR/Diastole interval ratio & Mean value of the interval ratios between diastole and RR in each heart beat \\
sd\_Ratio\_DiaRR & Std-dev RR/Diastole interval ratio & Std-dev value of the interval ratios between diastole and RR in each heart beat \\
m\_Ratio\_SysDia & Mean Systole/Diastole interval ratio & Mean value of the interval ratios between systole and diastole in each heart beat \\
sd\_Ratio\_SysDia & Std-dev Systole/Diastole interval ratio & Std-dev value of the interval ratios between systole and diastole in each heart beat \\
D{[}1-5{]}Shan & Detail coefficient shannon entropy & Shannon entropy of DWT detail coefficient 1-5 \\
A5Shan & Approximation coefficient shannon entropy & Shannon entropy of DWT approximation coefficient 1-5 \\
TotD{[}1-5{]}*Shan & Total detail coefficient shannon entropy & Total Shannon entropy of DWT detail coefficient 1-5 across signal \\
TotA5*Shan & Total approximation coefficient shannon entropy & Total Shannon entropy of DWT approximation coefficient 1-5 across signal \\ \hline
\end{tabulary}
\end{table}
\pagebreak
\subsection{Commandline Interface}\label{appendixB}