Icon

KN-302 Advanced Demand Forecasting Neural Networks v05

[KNIME Nodes] KN-302 Advanced Demand Forecasting Neural Networks
Step 1: Load the SalesForecating Training Data andTest Data, then clean the data ineach column. Step 3: Feature Enhancement.Account for the Year-Over-YearGrowth of Product Sales andscale sales to 2017 levels forreference. Deprecated. Step 2: Visualize the TrainingData and look for trends.Distinguish between Seasonalityand Year-Over-Year Growth. Step 5: Enhance Training Datafor the Neural Network. AddWeekday Encoding andSeasonality Encoding. Step 6: Group Training Data intopast 30-day samples + past 7-day samples for Training theNeural Network. Step 7: Generate and Train theNeural Networks. Compare arange of different Neural Networkmodels. Step 4: Normalize Sales TrainingData. Sum Sales by Item acrossall Stores. Shift to UnitDistribution(Mean = 0, SD = 1). KNIME Nodes (KN-302) - Advanced Demand Forecasting Neural NetworksProvides a Neural Network solution to the Kaggle Store Item Demand ForecastingChallenge. Input contains 5 years of store-item sales data. Deep Learning modelsuse both Keras Layer Nodes and DL Python Nodes. Output predicts the daily salesof 50 different items over the next two years. KNIME Installation Requirements:- KNIME Extension: KNIME Deep Learning - Keras Integration (Labs)- Python with Keras: https://www.knime.com/deeplearning#kerasData Source:Kaggle: Store Item Demand Forecasting Challengehttps://www.kaggle.com/c/demand-forecasting-kernels-only Step 8: Use the best NeuralNetwork model to predict futureSales. Step 9: Visualize the SalesPredictions. Check if modelpredicts day-of-week trends,seasonality trends, and year-over-year growth.DemandForecastingTraining DataCleanTest DataConvert Data TypeAdd Date FieldsRename ColumnsTraining DataTrendsScale Salesby Annual GrowthPeriodically EncodeMonth Input NodesMonth NodeEnergyItemDay Since StartSortInput to OutputMonth NodeEnergyPrepend Last30 Days ofTraining Data to2-Years of Test Data30 Day Input =1+12+7+31 = 51 columns7 Day Input =1+12+7+7 = 27 columnsTrain theNeural NetworksShift Sales toUnit DistributionDemandForecastingTest DataItem xTotal Scaled Sales xDay of YearOne-Hot EncodeDay of WeekWeekdayEncodingWeekdayEncodingPredict Next2-Years SalesVisualize Next2-Years SalesFile Reader Column Filter Clean Input Data VisualizeInput Data FeatureEnhancements SeasonalityEncoding Joiner Sorter Column Resorter Joiner Prepare Test Data PrepareTraining Data Neural Network Normalize Sales Test Data Normalize Test Weekday Encoding Joiner Joiner Make Predictions VisualizePredictions Step 1: Load the SalesForecating Training Data andTest Data, then clean the data ineach column. Step 3: Feature Enhancement.Account for the Year-Over-YearGrowth of Product Sales andscale sales to 2017 levels forreference. Deprecated. Step 2: Visualize the TrainingData and look for trends.Distinguish between Seasonalityand Year-Over-Year Growth. Step 5: Enhance Training Datafor the Neural Network. AddWeekday Encoding andSeasonality Encoding. Step 6: Group Training Data intopast 30-day samples + past 7-day samples for Training theNeural Network. Step 7: Generate and Train theNeural Networks. Compare arange of different Neural Networkmodels. Step 4: Normalize Sales TrainingData. Sum Sales by Item acrossall Stores. Shift to UnitDistribution(Mean = 0, SD = 1). KNIME Nodes (KN-302) - Advanced Demand Forecasting Neural NetworksProvides a Neural Network solution to the Kaggle Store Item Demand ForecastingChallenge. Input contains 5 years of store-item sales data. Deep Learning modelsuse both Keras Layer Nodes and DL Python Nodes. Output predicts the daily salesof 50 different items over the next two years. KNIME Installation Requirements:- KNIME Extension: KNIME Deep Learning - Keras Integration (Labs)- Python with Keras: https://www.knime.com/deeplearning#kerasData Source:Kaggle: Store Item Demand Forecasting Challengehttps://www.kaggle.com/c/demand-forecasting-kernels-only Step 8: Use the best NeuralNetwork model to predict futureSales. Step 9: Visualize the SalesPredictions. Check if modelpredicts day-of-week trends,seasonality trends, and year-over-year growth.DemandForecastingTraining DataCleanTest DataConvert Data TypeAdd Date FieldsRename ColumnsTraining DataTrendsScale Salesby Annual GrowthPeriodically EncodeMonth Input NodesMonth NodeEnergyItemDay Since StartSortInput to OutputMonth NodeEnergyPrepend Last30 Days ofTraining Data to2-Years of Test Data30 Day Input =1+12+7+31 = 51 columns7 Day Input =1+12+7+7 = 27 columnsTrain theNeural NetworksShift Sales toUnit DistributionDemandForecastingTest DataItem xTotal Scaled Sales xDay of YearOne-Hot EncodeDay of WeekWeekdayEncodingWeekdayEncodingPredict Next2-Years SalesVisualize Next2-Years SalesFile Reader Column Filter Clean Input Data VisualizeInput Data FeatureEnhancements SeasonalityEncoding Joiner Sorter Column Resorter Joiner Prepare Test Data PrepareTraining Data Neural Network Normalize Sales Test Data Normalize Test Weekday Encoding Joiner Joiner Make Predictions VisualizePredictions

Nodes

Extensions

Links

Download

To use this workflow in KNIME, download it from the below URL and open it in KNIME:

Download Workflow
Created by: Berlin
Created at: 2019-09-04
On NodePit since: 2021-02-05
Last update: 2024-12-21
Created with KNIME version: v4.3.0
Tags: Neural NetworkDeep LearningKaggle

Deploy, schedule, execute, and monitor your KNIME workflows locally, in the cloud or on-premises – with our brand new NodePit Runner.

Try NodePit Runner!