Icon

Energy Demand Prediction - Deployment

Simple auto-regressive model to predict a time series

- Simple means just raw data: no seasonality correction, stationarity assumption
- Auto means usage of past of the same time series for prediction. No other exogenous time series/data used.
- Regressive refers to the model: either a linear or a polynomial regression

We have new data up to now of energy consumption for groups of customers. We want to predict the next hours of energy consumption values.

We apply the previously trained regression model to the last 10 hour values to predict the 11th value; then we use this 11th predicted value to shape the past 10 value for the next prediction and we predict the 12th value; and so on ...

Deployment: Simple auto-regressive model to predict a time series

Con modelli salvati dal training:

Regressione Lineare o Polinomiale

Preparazione dei Dati

Pulizia

Loop per creare il nuovo passato

datetime to datetime type
String to Date&Time
Line Plot
tenere soloprediction
Column Filter
Estrarre una delletime series
Column Filter
per data crescente
Sorter
teniamo solo l'ultima riga di datila più recente
Row Filter
leggi i dati di deployment
CSV Reader
Recursive Loop End
Model Reader
Costruisce il passatoda x(t) a: x(t), x(t-1), x(t-2), ..., x(t-lag)lag =10
Lag Column
togliamo time seriesper non confonderci
Column Filter
Column Renamer
Regression Predictor
Recursive Loop Start
Column Resorter
chiamarla "time series"
Column Renamer
rimuoveretime series (-10)
Column Filter

Nodes

Extensions

Links