|
| 1 | +# Neuron class |
| 2 | +Neuron class provides LNU, QNU, RBF, MLP, MLP-ELM neurons learned with Gradient descent or LeLevenberg–Marquardt algorithm. This class is suitable for prediction on time series. |
| 3 | + |
| 4 | +# Dependencies |
| 5 | +Neuron class needs pandas and numpy to work propertly. |
| 6 | + |
| 7 | +# Example of usage |
| 8 | + |
| 9 | +Consider *Y* are targets and *X* are inputs. |
| 10 | + |
| 11 | +## LNUGD |
| 12 | +```python |
| 13 | +neuron = LNUGD() |
| 14 | +prediction = 1 |
| 15 | +yn, w, e, Wall, MSE = neuron.train(Y_train, X_train, epochs=2, prediction=prediction) |
| 16 | +yn, w, Wall, MSE, e = neuron.countSerie(Y, X, logging=False, prediction=prediction) |
| 17 | +``` |
| 18 | + |
| 19 | +## QNULM |
| 20 | +```python |
| 21 | +neuron = QNULM() |
| 22 | +prediction = 0 |
| 23 | +yn, w, e, Wall, MSE = neuron.train(Y_train, X_train, epochs=10, prediction=prediction) |
| 24 | +yn, w, MSE, e = neuron.countSerie(Y, X, logging=False, prediction=prediction) |
| 25 | +``` |
| 26 | + |
| 27 | +## RBF |
| 28 | +```python |
| 29 | +neuron = RBF() |
| 30 | +prediction = 1 |
| 31 | +neuron.train(Y_train, X_train, prediction=prediction) |
| 32 | +yn = neuron.count(Y,X, logging=True, beta=0.01, prediction=prediction) |
| 33 | +``` |
| 34 | +## MLPGD |
| 35 | +```python |
| 36 | +neuron = MLPGD() |
| 37 | +prediction = 0 |
| 38 | +yn = neuron.count(Y_train, X_train, prediction=prediction, epochs=5) |
| 39 | +yn = neuron.count(Y, X, prediction=prediction, epochs=1) |
| 40 | +``` |
| 41 | +## MLPELM |
| 42 | +```python |
| 43 | +neuron = MLPELM() |
| 44 | +prediction = 1 |
| 45 | +yn = neuron.count(Y_train, X_train, prediction = prediction, epochs = 10) |
| 46 | +yn = neuron.count(Y, X, prediction = prediction) |
| 47 | +``` |
| 48 | +## MLPLMWL |
| 49 | +```python |
| 50 | +neuron = MLPLMWL() |
| 51 | +prediction = 1 |
| 52 | +yn = neuron.count(Y, X, learningWindow = 50, overLearn = 10, prediction = prediction) |
| 53 | +``` |
0 commit comments