diff --git a/GLOSSARY.md b/GLOSSARY.md index e452821..675cd07 100644 --- a/GLOSSARY.md +++ b/GLOSSARY.md @@ -93,4 +93,40 @@ - **Steps25thAt**: Average time of day when 25% of steps was accumulated. - **Steps50thAt**: Average time of day when 50% of steps was accumulated. - **Steps75thAt**: Average time of day when 75% of steps was accumulated. - - **Steps95thAt**: Average time of day when 95% of steps was accumulated. \ No newline at end of file + - **Steps95thAt**: Average time of day when 95% of steps was accumulated. + + +## Random Forest Feature List +The table below describes the handcrafted features used as inputs to the Random Forest model. +These features are extracted from the Euclidean norm of the triaxial accelerometer data. + +| Feature Name | Description | Units | +|---------------------------------|--------------------------------------------------------------------|----------------| +| Moment features | +| avg | Mean | g | +| std | Standard deviation | g | +| skew | Skewness | | +| kurt | Kurtosis | | +| Quantile Features | +| min | Minimum | g | +| q25 | Lower quartile | g | +| med | Median | g | +| q75 | Upper quartile | g | +| max | Maximum | g | +| Autocorrelation features | +| acf_1st_max | Maximum autocorrelation | | +| acf_1st_max_loc | Location of 1st autocorrelation maximum | s | +| acf_1st_min | Minimum autocorrelation | | +| acf_1st_min_loc | Location of 1st autocorrelation minimum | s | +| acf_zeros | Number of autocorrelation zero-crossings | | +| Spectral features | +| pentropy | Signal's spectral entropy | nats | +| power | Signal's total power | g2/s | +| f1, f2, f3 | 1st, 2nd and 3rd dominant frequencies | Hz | +| p1, p2, p3 | Power spectral densities of respective dominant frequencies | g2/s | +| fft0, fft1, fft2, ... | Power spectral density for frequencies 0Hz, 1Hz, 2Hz, ... | g2/s | +| Peak features | +| npeaks | Number of peaks in the signal per second | 1/s | +| peaks_avg_promin | Average prominence of peaks | g | +| peaks_min_promin | Minimum prominence of peaks | g | +| peaks_max_promin | Maximum prominence of peaks | g | diff --git a/README.md b/README.md index a894b1a..760f8b5 100644 --- a/README.md +++ b/README.md @@ -116,7 +116,7 @@ $ stepcount sample.cwa -t rf When using the random forest model, a set of signal features is extracted from the accelerometer data. These features are subsequently used as inputs for the model's classification process. -For a comprehensive list of the extracted features, see [this list](https://github.com/OxWearables/stepcount/blob/main/rf-feature-list.md). +For a comprehensive list of the extracted features, see [the glossary](https://github.com/OxWearables/stepcount/blob/main/GLOSSARY.md#random-forest-feature-list). ### Crude vs. Adjusted Estimates diff --git a/rf-feature-list.md b/rf-feature-list.md deleted file mode 100644 index 902e151..0000000 --- a/rf-feature-list.md +++ /dev/null @@ -1,34 +0,0 @@ -# Random Forest Feature List -The table below presents an overview of the handcrafted features that have been curated as inputs for the Random Forest model. -These features are extracted from the Euclidean norm of the triaxial accelerometer data, minus 1 to remove gravity. - -| Feature Name | Description | Units | -|---------------------------------|--------------------------------------------------------------------|----------------| -| Moment features | -| avg | Mean | g | -| std | Standard deviation | g | -| skew | Skewness | | -| kurt | Kurtosis | | -| Quantile Features | -| min | Minimum | g | -| q25 | Lower quartile | g | -| med | Median | g | -| q75 | Upper quartile | g | -| max | Maximum | g | -| Autocorrelation features | -| acf_1st_max | Maximum autocorrelation | | -| acf_1st_max_loc | Location of 1st autocorrelation maximum | s | -| acf_1st_min | Minimum autocorrelation | | -| acf_1st_min_loc | Location of 1st autocorrelation minimum | s | -| acf_zeros | Number of autocorrelation zero-crossings | | -| Spectral features | -| pentropy | Signal's spectral entropy | nats | -| power | Signal's total power | g2/s | -| f1, f2, f3 | 1st, 2nd and 3rd dominant frequencies | Hz | -| p1, p2, p3 | Power spectral densities of respective dominant frequencies | g2/s | -| fft0, fft1, fft2, ... | Power spectral density for frequencies 0Hz, 1Hz, 2Hz, ... | g2/s | -| Peak features | -| npeaks | Number of peaks in the signal per second | 1/s | -| peaks_avg_promin | Average prominence of peaks | g | -| peaks_min_promin | Minimum prominence of peaks | g | -| peaks_max_promin | Maximum prominence of peaks | g |