Python code to read inputs from Excel file, predict outputs using NN and write outputs back to Excel

This is for non-business use. I have multiple similar Excel files. They have currently 9 input columns (side by side and in same position) and 1-3 output columns. Python needs to read these inputs and predict missing outputs and write them to Excel. I can install whatever you need for Python (I have Anaconda already installed), but I prefer Keras and TensorFlow. Requirements: - Read certain Excel file from computer (xlsx). Decimal separators are commas. - Use certain range of rows and columns (variables) of Excel as training and testing inputs and outputs - Split the data randomly to training and testing sets - Ability to tune the neural network (multiple layers, number of neurons per layer, include all hyperparameters (epochs, optimizer, learning rate and momentum, etc)) o Include grid search with printed output for faster neural network optimization - Ability to select number of outputs for prediction (1 to 3 columns) (this isn't mandatory: 1 is enough, 3 is preferred) - Automatically selects the output rows (find the first empty row for certain column in Excel and write from there till the end) - Outputs should be non-binary and decimal separator needs to be comma (for example 0,84234) - Write the outputs to Excel - Ability to save network (multiple names for different Excel-files) and load the saved network for endless reuse