make-prediction

---

Introduction

---

A CLI tool for making predictions using a pre-trained model. This tool is designed to be simple and efficient, allowing users to quickly generate predictions without needing to modify the underlying code.

Motivation

---

This project is part of my learning journey to explore suitable models for predicting cryptocurrency prices. It also provides midterm reporting results for my Data Mining course.

Prerequisites

---

• Python 3.x
• Any necessary dependencies. See the requirements.txt file for a list of required packages.

Installing TA-Lib

---

TA-Lib requires special installation steps:

On Windows

---

Download and install the wheel from unofficial Windows binaries

On macOS

---

brew install ta-lib
pip install ta-lib

On Linux

---

wget http://prdownloads.sourceforge.net/ta-lib/ta-lib-0.4.0-src.tar.gz
tar -xzf ta-lib-0.4.0-src.tar.gz
cd ta-lib/
./configure --prefix=/usr
make
sudo make install
pip install ta-lib

Installation

---

Clone the repository and install the required packages. After that, run the following command to install the package in editable mode:

pip install -e .

Usage

---

Make Prediction CLI
====================

Available commands:
- help     : Show this help message.
- predict  : Run prediction using trained model.
- new      : Initialize a new project setup.

First, create your config.ini and fill in the required fields. You can use the provided config.ini.example as a template. Then use the new command to create a new model with your configuration file.

predictor new config.ini

After training the model, you can use the predict command to make predictions.

predictor predict config.ini

Configuration File

---

The configuration file is a .ini file that contains the following sections:

[CONFIG]
model = lstm # Model type: lstm, gru, bi_lstm
mode = simple # Mode: simple, complex
window_size = 30
symbol = BTC
start = Jun 01 2018
end = Dec 31 2023
scaler_path = ./.scaler/scaler.pkl
model_path = ./.model/model.keras

Reference

---

For more details, refer to the paper:

• A Novel Cryptocurrency Price Prediction Model Using GRU, LSTM and bi-LSTM Machine Learning Algorithms