tsg-lib 0.1.3

A modular library for synthetic time series generation.

tsg – Modular Time Series Generator Library

tsg is a lightweight Python library for generating synthetic time series data using modular and composable generators.
It is designed for research, simulation, and testing of sequential decision-making algorithms, especially in finance and machine learning.

---

🚀 Features

• 📈 Core generators: linear trend, periodic trend, constant value
• 🎲 Noise wrappers: add Gaussian noise or build your own modifiers
• 🔁 Stateful generators with reset() support
• 🧱 Easy to extend with your own generator or modifier classes

---

📦 Installation

Install through GitHub from any location:

pip install git+https://github.com/MSCA-DN-Digital-Finance/tsg.git

Install through PyPi from any location:

pip install tsg-lib

🛠️ Example Usage

Here’s how to generate a noisy linear trend time series using tsg:

from tsg.generators import LinearTrendGenerator
from tsg.modifiers import GaussianNoise

# Create a linear trend generator (increasing by +1 each step)
base_generator = LinearTrendGenerator(start_value=100, up=True)

# Wrap it with Gaussian noise (mean=0, std=1)
noisy_generator = GaussianNoise(base_generator, mu=0.0, sigma=1.0)

# Generate a few data points
values = []
for _ in range(10):
    value = noisy_generator.generate_value(None)
    values.append(value)

print(values)

🧠 API Overview

Core Generators (tsg.generators)

Class	Description
LinearTrendGenerator	Linearly increases or decreases the value at each step
ConstantGenerator	Returns a fixed value (e.g., simulates cash)
PeriodicTrendGenerator	Generates a sinusoidal time series with set amplitude and frequency

Modifier Wrappers (tsg.modifiers)

Class	Description
GaussianNoise	Adds Gaussian noise (N(mu, sigma)) to any base generator

All components implement the BaseGenerator interface with:

• generate_value(last_value) – returns the next value in the sequence
• reset() – resets any internal state (optional for stateless generators)

---

Acknowledgments

Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or European Research Executive Agency (REA). Neither the European Union nor the granting authority can be held responsible for them.

License

MIT — see LICENSE.