This is a project about collecting data in 4 different movement to perform machine learning and neural network.
I chose running, walking, jumping and bend_knee during the moment of collecting my body data. The data are collected individually by its group. Also, there is another data was collected by combining all movement of myself.
I chose KNeighborsClassifier, because this machine learning is classification model. Also, KNN has the property of find how spread are the data surrounding by the K-value. Besides, using grid search can let me to pick my k-value, weight and display the accuracy.
In the end, the value of count, mean, std, min, 25%, 50%, 75% and max will be display in a table format. Also, finding the correlation
Here is an example of the testing data result
Example of autocorrelation plotfunction
Example of the result of training data
In this neural network, I used simpleRNN, GRU, LSTM with different layers. Below are the results for LSTM and simpleRNN for 3 different datasets.
traning_data_sets
| neurons | layers | time steps | accuracy |
|---|---|---|---|
| LSTM | 2 | 10 | 0.5993 |
| LSTM | 2 | 20 | 0.5431 |
| LSTM | 2 | 40 | 0.4195 |
| LSTM | 3 | 10 | 0.8436 |
| LSTM | 3 | 20 | 0.8071 |
| LSTM | 3 | 40 | 0.6404 |
| LSTM | 4 | 10 | 0.8146 |
| LSTM | 4 | 20 | 0.6592 |
| LSTM | 4 | 40 | 0.5094 |
| SimpleRNN | 2 | 10 | 0.2603 |
| SimpleRNN | 2 | 20 | 0.4307 |
| SimpleRNN | 2 | 40 | 0.3708 |
| SimpleRNN | 3 | 10 | 0.4522 |
| SimpleRNN | 3 | 20 | 0.3352 |
| SimpleRNN | 3 | 40 | 0.4644 |
| SimpleRNN | 4 | 10 | 0.265 |
| SimpleRNN | 4 | 20 | 0.2622 |
| SimpleRNN | 4 | 40 | 0.2547 |
testing_data_sets
| neurons | layers | time steps | accuracy |
|---|---|---|---|
| LSTM | 2 | 10 | 0.2731 |
| LSTM | 2 | 20 | 0.3981 |
| LSTM | 2 | 40 | 0.2407 |
| LSTM | 3 | 10 | 0.6343 |
| LSTM | 3 | 20 | 0.5093 |
| LSTM | 3 | 40 | 0.444 |
| LSTM | 4 | 10 | 0.287 |
| LSTM | 4 | 20 | 0.2685 |
| LSTM | 4 | 40 | 0.2593 |
| SimpleRNN | 2 | 10 | 0.3056 |
| SimpleRNN | 2 | 20 | 0.2407 |
| SimpleRNN | 2 | 40 | 0.3333 |
| SimpleRNN | 3 | 10 | 0.287 |
| SimpleRNN | 3 | 20 | 0.3519 |
| SimpleRNN | 3 | 40 | 0.2593 |
| SimpleRNN | 4 | 10 | 0.287 |
| SimpleRNN | 4 | 20 | 0.2685 |
| SimpleRNN | 4 | 40 | 0.2593 |
Moments are done continuously
| neurons | layers | time steps | accuracy |
|---|---|---|---|
| LSTM | 2 | 10 | 0.1943 |
| LSTM | 2 | 20 | 0.2759 |
| LSTM | 2 | 40 | 0.4419 |
| LSTM | 3 | 10 | 0.6 |
| LSTM | 3 | 20 | 0.4318 |
| LSTM | 3 | 40 | 0.4186 |
| LSTM | 4 | 10 | 0.3314 |
| LSTM | 4 | 20 | 0.3333 |
| LSTM | 4 | 40 | 0.3953 |
| SimpleRNN | 2 | 10 | 0.0914 |
| SimpleRNN | 2 | 20 | 0.1034 |
| SimpleRNN | 2 | 40 | 0.2093 |
| SimpleRNN | 3 | 10 | 0.3714 |
| SimpleRNN | 3 | 20 | 0.3563 |
| SimpleRNN | 3 | 40 | 0.4884 |
| SimpleRNN | 4 | 10 | 0.3314 |
| SimpleRNN | 4 | 20 | 0.3333 |
| SimpleRNN | 4 | 40 | 0.3256 |