ME435Lab3

.pdf

School

Pennsylvania State University *

*We aren’t endorsed by this school

Course

435

Subject

Economics

Date

Apr 3, 2024

Type

pdf

Pages

3

Uploaded by PENNY.2

Report
Q1: If we are planning to use two features to identify the Brush teeth motion, which two features could we choose? Explain your choices. Hint: Change the predictors combination next to the Scatter Plot. Answer: Average acceleration Z and average resultant acceleration because it creates the largest gap between the brush teeth values and the other activities while keeping them somewhat clustered. This is shown in the graph below where brush teeth is blue. Original data set: reference_table o L5 —@ [ o o ° - 2 S 11} ® [ D » © g ° g . o °n £ e ® o % 5105— ¥ o0 © 3 g . Ty d > @ e ° ® * @ e @ ® 9 5} » & ° ° o ° e %%, z .‘o o ® H e g0 e 0. o o o . o © 1 -no..". L L @ ° %% ® @ & .\ = - e % o‘.fi’ ° o 0.95 i i i i i i 0 02 04 06 08 1 AverageAccelerationZ Q2: Repeat the above steps with the two features you chosen to track the Brush teeth motion in question 1. Compare at least three models by showing their confusion matrix and explain which one gives the best prediction for separating the Brush teeth movement from other activities. The best model is the fine KNN since it has no false positives and a 100% true positive detection rate for our data. Fine Tree is significantly worse because it miscategorizes brushing teeth as drinking water a third of the time. Cosine KNN on the other hand, miscategorizes drinking water as brushing teeth. Fine KNN Model 3.4 Brush_teeth Drink_glass H33% 26.7% Liedown_bed 66.7% | 33.3% True Class Pour_water 733% | 26.7% Walk SN 3.3% Brush_teet®rink_gladsedown_beBour_water Walk TPR FNR Predicted Class Fine Tree
Model 3.1 Brush_teeth | 186.7% | 33.3% Drink_glass 63.3% | 36.7% 12 0w s ?, Liedown_bed 66.7% | 33.3% e }.— Pour_water 70.0% | 30.0% Walk eagx 6.7% Brush_teetBrink_gladsedown_beBour_water Walk TPR FNR Predicted Class Cosine KNN Model 3.7 &, Brush_teeth [N66:7% 33.3% ,-_66 1% | 33.3% Drink_glass 0.0% 30.0% 1 70.0% | 30.0% . } 0 = . g Liedown_bed | 33.3% 1.1% | 556% . 100.0% 3 = Pour_water 50.0% 50.0% 50.0% Walk 00.0 100.0% Brush_teet®rink_gladsedown_beBour_water Walk TPR FNR Predicted Class Q3: Train your model by adding and changing features, applying different model type and/or validation method. Our goal is to obtain a model which will provide at least 75% true predicition for all five activites. Summarize your choice, attach and explain the confusion matrix result. Students must complete this task by using no more than 5 features. We used average acceleration x, average acceleration y, average acceleration z, average absolute difference x, and average peak to peak time x. The three average accelerations provide a good basis for accelerations in the motion, but the absolute difference and the time give a good idea of how each one varies.
Your preview ends here
Eager to read complete document? Join bartleby learn and gain access to the full version
  • Access to all documents
  • Unlimited textbook solutions
  • 24/7 expert homework help