Supervised Learning Problem
This is a simple example workflow for multivariant time series analysis using an LSTM based recurrent neural network and implemented via the KNIME Deep Learning - Keras Integration. It is based on the bike demand prediction dataset from Kaggle and trains a model to predict the demand in the next hour based on the demand and the other features in the last 10 hours.
RNN architecture type = many-to-one
target column = "cnt'
features column = all column (except "timestamp"
timestamp column = "timestamp"
notes:
this workflow is expanded version of "Multivariate Time Series Analysis with an RNN" from Community Hub with these environment specification:
rewrite on: May 2025
KNIME version: 5.4
The Dataset
London Bike Sharing dataset have purpose is to try predict the future bike shares.
Metadata:
"timestamp" - timestamp field for grouping the data
"cnt" - the count of a new bike shares
"t1" - real temperature in C
"t2" - temperature in C "feels like"
"hum" - humidity in percentage
"wind_speed" - wind speed in km/h
"weather_code" - category of the weather
"is_holiday" - boolean field - 1 holiday / 0 non holiday
"is_weekend" - boolean field - 1 if the day is weekend
"season" - category field meteorological seasons:
“Season” category description:
· 0 = spring ;
· 1 = summer;
· 2 = fall;
· 3 = winter.
"weather_code" category description:
· 1 = Clear ; mostly clear but have some values with haze/fog/patches of fog/ fog in vicinity
· 2 = scattered clouds / few clouds
· 3 = Broken clouds
· 4 = Cloudy
· 7 = Rain/ light Rain shower/ Light rain
· 10 = rain with thunderstorm
· 26 = snowfall
94 = Freezing Fog