Scribd
Upload
14 views7 pages

Loadalgarve MLP

The document discusses loading time series data from a MATLAB file into a Pandas dataframe. It then cleans the data, splits it into training and test sets, and uses a multi-layer perceptron model to perform time series forecasting on the data. The model is trained over 40 epochs with decreasing loss and increasing accuracy over time.

Uploaded by

Rony Rock
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
14 views7 pages

Loadalgarve MLP

The document discusses loading time series data from a MATLAB file into a Pandas dataframe. It then cleans the data, splits it into training and test sets, and uses a multi-layer perceptron model to perform time series forecasting on the data. The model is trained over 40 epochs with decreasing loss and increasing accuracy over time.

Uploaded by

Rony Rock
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

loadalgarve MLP

March 24, 2024

[ ]: import numpy as np
import pandas as pd
from scipy.io import loadmat

'''Constructing a pandas dataframe from a matlab .mat file'''

data_set = loadmat("LCAlgarve.mat")
mat = {k: v for k, v in data_set.items() if k[0] != '_'}
data = pd.DataFrame({k: pd.Series(v[:, 0]) for k, v in mat.items()})
data.to_csv("example.csv")

[ ]: df = pd.read_csv('example.csv')
df.dtypes

[ ]: Unnamed: 0 int64
DayCode1 float64
DayCode2 float64
Occupation1 float64
Occupation2 int64
Power1 float64
Power2 float64
dtype: object

[ ]: df.shape

[ ]: (1751, 7)

[ ]: df.isnull().any()

[ ]: Unnamed: 0 False
DayCode1 True
DayCode2 False
Occupation1 True
Occupation2 False
Power1 True
Power2 False
dtype: bool

1
[ ]: df.head()

[ ]: Unnamed: 0 DayCode1 DayCode2 Occupation1 Occupation2 Power1 \


0 0 0.5 0.8 3.0 5 0.755691
1 1 0.5 0.8 3.0 5 0.339651
2 2 0.5 0.8 3.0 5 0.222168
3 3 0.5 0.8 3.0 5 0.404500
4 4 0.5 0.8 3.0 5 0.956347

Power2
0 0.747135
1 1.370311
2 0.759630
3 1.125687
4 1.910926

[ ]: df.isnull().any()

[ ]: Unnamed: 0 False
DayCode1 True
DayCode2 False
Occupation1 True
Occupation2 False
Power1 True
Power2 False
dtype: bool

[ ]: df.duplicated()

[ ]: 0 False
1 False
2 False
3 False
4 False
…
1746 False
1747 False
1748 False
1749 False
1750 False
Length: 1751, dtype: bool

[ ]: data1 = pd.read_csv('example.csv', index_col=0)


print(data1.dtypes)

DayCode1 float64
DayCode2 float64
Occupation1 float64

2
Occupation2 int64
Power1 float64
Power2 float64
dtype: object

[ ]: import sklearn
from sklearn import preprocessing
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.neural_network import MLPClassifier
from sklearn.metrics import classification_report, confusion_matrix

data1 = pd.read_csv('example.csv', index_col=0)


data2 = data1[['DayCode2', 'Occupation2', 'Power2']]
print(data2)

DayCode2 Occupation2 Power2


0 0.80 5 0.747135
1 0.80 5 1.370311
2 0.80 5 0.759630
3 0.80 5 1.125687
4 0.80 5 1.910926
… … … …
1746 0.30 3 2.672848
1747 0.30 3 2.337520
1748 0.30 3 2.354926
1749 0.35 2 1.933089
1750 0.35 2 2.036769

[1751 rows x 3 columns]

[ ]: # https://gist.githubusercontent.com/rupak-roy/edd1afa3d07bb928f70df5a344a08c6a/
↪raw/20dfe5f218815ad28e60db1d065e009045f3e0fa/

↪Multi-Layer%20Perception%20(MLP)%20for%20time%20series%20forecasting

dataset = data2.values
data = dataset.flatten()
raw_seq = data.tolist()

[ ]: # split a univariate sequence into samples


def split_sequence(sequence, n_steps):
X, y = list(), list()
for i in range(len(sequence)):
# find the end of this pattern
end_ix = i + n_steps
# check if we are beyond the sequence
if end_ix > len(sequence)-1:
break
# input and output sample

3
seq_x, seq_y = sequence[i:end_ix], sequence[end_ix]
X.append(seq_x)
y.append(seq_y)
return array(X), array(y)

[ ]: from numpy import array


from keras.models import Sequential
from keras.layers import Dense

# choose a number of time steps


n_steps = 3
# split into samples
X, y = split_sequence(raw_seq, n_steps)

# define model
model = Sequential()
model.add(Dense(150, activation='relu', input_dim=n_steps))
model.add(Dense(1))
model.compile(optimizer='adam', loss='mse', metrics =['accuracy'])

C:\Users\iamro\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n
2kfra8p0\LocalCache\local-packages\Python311\site-
packages\keras\src\layers\core\dense.py:88: UserWarning: Do not pass an
`input_shape`/`input_dim` argument to a layer. When using Sequential models,
prefer using an `Input(shape)` object as the first layer in the model instead.
super().__init__(activity_regularizer=activity_regularizer, **kwargs)

[ ]: # fit model

history=model.fit(X, y, epochs=40, verbose=1)

Epoch 1/40
165/165 �������������������� 1s 943us/step -
accuracy: 0.0552 - loss: 0.9641
Epoch 2/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0526 - loss: 0.1490
Epoch 3/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0535 - loss: 0.1530
Epoch 4/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0500 - loss: 0.1494
Epoch 5/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0551 - loss: 0.1623
Epoch 6/40

4
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0558 - loss: 0.1314
Epoch 7/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0516 - loss: 0.1423
Epoch 8/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0514 - loss: 0.1502
Epoch 9/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0533 - loss: 0.1519
Epoch 10/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0502 - loss: 0.1419
Epoch 11/40
165/165 �������������������� 0s 964us/step -
accuracy: 0.0518 - loss: 0.1426
Epoch 12/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0503 - loss: 0.1524
Epoch 13/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0518 - loss: 0.1734
Epoch 14/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0537 - loss: 0.1584
Epoch 15/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0563 - loss: 0.1523
Epoch 16/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0548 - loss: 0.1581
Epoch 17/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0540 - loss: 0.1571
Epoch 18/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0594 - loss: 0.1407
Epoch 19/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0548 - loss: 0.1636
Epoch 20/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0634 - loss: 0.1570
Epoch 21/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0587 - loss: 0.1582
Epoch 22/40

5
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0467 - loss: 0.1653
Epoch 23/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0564 - loss: 0.1517
Epoch 24/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0584 - loss: 0.1477
Epoch 25/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0610 - loss: 0.1595
Epoch 26/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0521 - loss: 0.1462
Epoch 27/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0508 - loss: 0.1465
Epoch 28/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0550 - loss: 0.1374
Epoch 29/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0515 - loss: 0.1387
Epoch 30/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0588 - loss: 0.1497
Epoch 31/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0568 - loss: 0.1446
Epoch 32/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0557 - loss: 0.1574
Epoch 33/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0518 - loss: 0.1488
Epoch 34/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0504 - loss: 0.1515
Epoch 35/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0552 - loss: 0.1662
Epoch 36/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0524 - loss: 0.1371
Epoch 37/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0539 - loss: 0.1671
Epoch 38/40

6
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0544 - loss: 0.1384
Epoch 39/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0488 - loss: 0.1528
Epoch 40/40
165/165 �������������������� 0s 1ms/step -
accuracy: 0.0518 - loss: 0.1438

[ ]: from matplotlib import pyplot


pyplot.plot(history.history['loss'], label='train')
pyplot.plot(history.history['accuracy'], label='test')
pyplot.legend()
pyplot.show()

You might also like