Resumen

The Shiny application to be developed is based on the final project (1) presented in machine learning course. This project is to implement a model to predict what the best routine to perform an exercise with weights.








(1) url Project used machine learning [final] (http://htmlpreview.github.io/?https://github.com/magzupao/machine_learning/blob/master/project_machine_learning_magz.html)

(2) Resources Final Project: repository

Modelo

Our initial model consists of 5 measurements recorded by sensors placed; wrist, forearm and arm several people. Where A is better and E is the worst routine.

Classification:
A
B
C
D
E

It was recorded for each user samples were grouped in a matrix composed of data of 160 variables and 19622 records.

Process

After processing (data cleaning) the model, we have a data matrix consisting of 53 variables and records 19622. where we have:

52 -> variables predictors
1 -> variable discriminant


Based on the analysis model where we have quantitative and qualitative variables can solve a case prediction algorithm using the Random Forest.

Results

Our model with a value of tree = 30 —> 100 is stable, and has a very low tracking error, which is reliable.

## 
## Call:
##  randomForest(formula = classe ~ ., data = new_df_training, ntree = 30,      mtry = 6, replace = T) 
##                Type of random forest: classification
##                      Number of trees: 30
## No. of variables tried at each split: 6
## 
##         OOB estimate of  error rate: 0.68%
## Confusion matrix:
##      A    B    C    D    E class.error
## A 5572    7    0    1    0 0.001433692
## B   24 3754   16    0    3 0.011324730
## C    1   19 3395    7    0 0.007890123
## D    3    3   36 3172    2 0.013681592
## E    0    1    3    7 3596 0.003049626

.