Prepara i dati per il machine learning

Analizzare i dati IoT in Python

Matthias Voppichler

IT Developer

Ripasso di machine learning

Apprendimento supervisionato
- Classificazione
- Regressione
Apprendimento non supervisionato
- Analisi di cluster
Deep learning
- Reti neurali

Ripasso di machine learning

Apprendimento supervisionato
- Classificazione
- Regressione
Apprendimento non supervisionato
- Analisi di cluster
Deep learning
- Reti neurali

Label

print(environment_labeled.head())

                     humidity  temperature  pressure   label
timestamp                                                   
2018-10-01 00:00:00      81.0         11.8    1013.4       1
2018-10-01 00:15:00      79.7         11.9    1013.1       1
2018-10-01 00:30:00      81.0         12.1    1013.0       1
2018-10-01 00:45:00      79.7         11.7    1012.7       1
2018-10-01 01:00:00      84.3         11.2    1012.6       1

Train / Test split

Suddividere le serie temporali

Il modello non deve vedere il test durante l’addestramento
Niente split casuale
Il modello non deve guardare nel futuro

Train/test split

split_day = "2018-10-13"

train = environment[:split_day]
test = environment[split_day:]

print(train.iloc[0].name)
print(train.iloc[-1].name)
print(test.iloc[0].name)
print(test.iloc[-1].name)

2018-10-01 00:00:00
2018-10-13 23:45:00
2018-10-14 00:00:00
2018-10-15 23:45:00

Suddivisione train/test

Feature e label

X_train = train.drop("target", axis=1)
y_train = train["target"]
X_test = test.drop("target", axis=1)
y_test = test["target"]

print(X_train.shape)
print(y_train.shape)

(1248, 3)
(1248,)

Regressione logistica

from sklearn.linear_model import LogisticRegression

logreg = LogisticRegression()

logreg.fit(X_train, y_train)

print(logreg.predict(X_test))

[0 0 1 1 1 1 1 0 0]

Passons à la pratique !

Analizzare i dati IoT in Python