IT1244 Project README

Packages

Additional Packages Required:
pyarrow 14.0.0
fastparquet 2023.10.1
pandas 2.1.2
scikit-learn 1.3.2
tensorflow 2.14.0
numpy 1.26.1

You can install the packages using: pip install <package_name>
Or upgrade using: pip install <package_name> --upgrade

How to get started (LSTM)

1. Ensure the required packages are installed and of the correct version.

2. Place the data.parquet file in the same directory as the main_lstm.ipynp jupyter file.

3. If using pretrained model(s), place the saved_models/ folder in the same directory as the main_lstm.ipynp jupyter file. Some pretrained models are provided. More information can be found in Code Parameters.

4. Set the parameters to obtain the desired results within the main_lstm.ipynp jupyter file as shown in Code Parameters.

5. Run the program.

Code Parameters

The following parameters are located at the top of the main jupyter file.

• Setting is_train_model to True will train a new model. (Overwrites existing model(s))

  Setting is_train_model to False will use pretrained models in the saved_models/ folder.

• run_option has 4 options:

  • 1 will print relative root-mean-square-error for open, high, low, close for a single company based on a company-specific model.

  • 2 will print relative root-mean-square-error for open, high, low, close for a single company based on a sector-specific model.

  • 3 will print an average relative root-mean-square error for open, high, low, close for all companies in a specified sector based on a company-specific model.

  • 4 will print an average relative root-mean-square error for open, high, low, close for all companies in a specified sector based on a sector-specific model.

• COMPANY_INDEX selects the company to be used, ranges from 0 to 492.

  • Affects run_option 1.

• SECTOR_INDEX selects the sector to be used, ranges from 0 to 10.

  • Affects run_option 2,3 and 4.
  • Pretrained models for run_option 3 and 4 for sectors 0, 1, 2, 3, 4, 9 and 10 are provided.

How to get started for secondary models (Linear Regression, PCR)

1. Ensure the required packages are installed and of the correct version.

2. Place the data.parquet file in the same directory as the Other_Models.ipynp jupyter file.

3. Run the program.

How to get started for feature visualisation (EDA)

1. Ensure the required packages are installed and of the correct version.

2. Place the data.parquet file in the same directory as the EDA.ipynp jupyter file.

3. Run the program.