Review_for_Final.md

Review for Final

Fall 2019

The final will focus on data analysis, basic model training, code correction, and model testing. (I will continue to expand on this document)

It will be an online format to be submitted through Canvas. 

You will need access to Quandl for data for the final. https://github.com/Machine-Learning-for-Finance/Machine-Learning-Algorithms/blob/master/01-Data%20Loading/Quandl.ipynb

Google Colab should be used to run any of the examples. https://colab.research.google.com/notebooks/welcome.ipynb#recent=true

ALL code used for the questions must be provided.

Data Analysis:

• Determine class distribution.
  • https://github.com/Machine-Learning-for-Finance/Machine-Learning-Algorithms/blob/master/02-Data%20Analysis/02-Entire_Acquisitions_File.ipynb
  • https://github.com/Machine-Learning-for-Finance/Machine-Learning-Algorithms/blob/master/02-Data%20Analysis/03-Combining%20Datasets.ipynb
• Determine class weightings.
• Determine descriptive statistics.
  • https://github.com/Machine-Learning-for-Finance/Machine-Learning-Algorithms/blob/master/02-Data%20Analysis/01-Basic%20Analysis.ipynb Basic Model Training:
• Be able to train
  • a classifier network,
    • https://github.com/Machine-Learning-for-Finance/Machine-Learning-Algorithms/blob/master/04-Model%20Creation/04-Initial_Model.ipynb
  • regressor network, and
    • https://github.com/Machine-Learning-for-Finance/Machine-Learning-Algorithms/blob/master/04-Model%20Creation/06-Regression.ipynb
  • autoencoder network.
    • https://github.com/Machine-Learning-for-Finance/Machine-Learning-Algorithms/blob/master/01-Data%20Loading/RegimeDetection.ipynb

Be able to use all three models to predict values. Code Correction:

• Given Python code for either: data normalization, training, or usage of a model, determine any bugs with the code and make the necessary corrections. Model Testing:
• Given access to a model through a REST API (example code of how to make a call to it will be given) test the model on a provided dataset and determine it's accuracy and any classes that it performs poorly on.