Python: Feature Extraction, Selection, Parameter and Hyperparameter Optimization

Provided a small 778Kb data set and a sample RapidMiner .rmp file: In Python: #Convert Date field to ISO 8601 format; and declare Date as ID (key) #Generate Classification Target Variable(Label, Binomial Classification '1'= up; '0' = down). #Select and Extract Features; Apply Multi Objective Feature Selection to reduce attribute set/maximize performance #Apply Gradient Boosted Tree for Parameter Optimization #Apply 5-day Windowing Multi-variate Forecasting Analysis. Step size 1 #Apply Sliding Window to backtest Window: 10 sample window-width; Step size: -1, Horizon: -1 #Apply SVM with RBF Kernel for Time Series Forecasting #Optimize C and Gamma parameters #Optimize Parameters (apply Walk Forward) Outputs in .xlsx format: 1. Generate List Features ranked by correlation and covariance 2. Generate List of Parameters 3. Generate List of Hyperparameters 4. Generate Cross Validation Metrics: Apply Wrapper Cross-validation on Selected Features. Accuracy, Confusion Matrix performance metrics. 5. Generate Training Data set (with derivative labels) 6. Generate 2 Out of Sample Test Data set (With Labels and without Labels) Random 20% of population. Provide Script in .ipynb format (Jupyter) The output (Training and Testing Dataset) Performance Validation: Cross Validation achieves Accuracy at 90% or greater when run through our Validation model. Potential add on work: Convert Label from nominal to numerical measuring magnitude (+/-)