LSTM - Time Series Forecasting

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