Predict hand angles with expert rules based on observed values.
This model was created from observed values in a corpus. - Collected data: 3769 values - Filtered data: 3722 values (outliers removed)
Statistics by Hand Position - b : Mean = 64.29, Std Dev = 7.87 - c : Mean = 61.34, Std Dev = 7.37 - m : Mean = 62.58, Std Dev = 8.12 - s : Mean = 68.49, Std Dev = 11.07 - t : Mean = 54.02, Std Dev = 9.96 => The position is strongly correlated to the angle.
Statistics by Hand Shape - 1 : Mean = 63.37, Std Dev = 11.85 - 2 : Mean = 64.03, Std Dev = 10.54 - 3 : Mean = 65.73, Std Dev = 10.89 - 4 : Mean = 66.25, Std Dev = 11.50 - 5 : Mean = 62.38, Std Dev = 10.01 - 6 : Mean = 65.04, Std Dev = 12.08 - 7 : Mean = 60.59, Std Dev = 11.59 - 8 : Mean = 60.98, Std Dev = 9.50 => The shape has no impact on the angle value.
Statistics by Speaker - AM: Mean = 63.86, Std Dev = 12.91 - CH: Mean = 73.80, Std Dev = 10.41 - LM: Mean = 61.72, Std Dev = 7.35 - ML: Mean = 56.81, Std Dev = 5.88 - VT: Mean = 65.60, Std Dev = 11.10 => There is a speaker effect, but it should be accounted for to create a general model.
Statistics by Condition - syll: Mean = 61.17, Std Dev = 8.68 - word: Mean = 58.30, Std Dev = 12.72 - sent: Mean = 64.88, Std Dev = 10.05 - text: Mean = 66.11, Std Dev = 10.49 => The condition affects the angle, possibly by influencing movement precision; so it should be accounted for to create a general model.
