Found in the #16 correctness/prose/comments audit (verified).
The standard-deviation feature is given a wrong column name in both gesture notebooks. The value computed is np.std(trial.accel.mag) (std of the magnitude signal), but the name appended is "std(mag.max())":
features.append(np.std(trial.accel.mag))
feature_names.append("std(mag.max())") # <-- should be "std(mag)"
mag.max() is a single scalar (its std would be 0/NaN), so the name is simply wrong and misleads every feature table, correlation heatmap, and classification report that displays it.
Locations
Projects/GestureRecognizer/GestureRecognizer-FeatureBased.ipynb (the extract_features_from_gesture_set cell)
Projects/GestureRecognizer/Feature Selection and Hyperparameter Tuning.ipynb (its copy of the same function)
Fix
Rename to "std(mag)" (or "mag.std()") in both.
Found in the #16 correctness/prose/comments audit (verified).
The standard-deviation feature is given a wrong column name in both gesture notebooks. The value computed is
np.std(trial.accel.mag)(std of the magnitude signal), but the name appended is"std(mag.max())":mag.max()is a single scalar (its std would be 0/NaN), so the name is simply wrong and misleads every feature table, correlation heatmap, and classification report that displays it.Locations
Projects/GestureRecognizer/GestureRecognizer-FeatureBased.ipynb(theextract_features_from_gesture_setcell)Projects/GestureRecognizer/Feature Selection and Hyperparameter Tuning.ipynb(its copy of the same function)Fix
Rename to
"std(mag)"(or"mag.std()") in both.