https://github.com/bheemisme/time-series-forecasting

Time series forecasting on multiple datasets
https://github.com/bheemisme/time-series-forecasting

lstm machine-learning python scikit-learn time-series-forecasting xgboost

Last synced: about 2 months ago
JSON representation

Time series forecasting on multiple datasets

• Host: GitHub
• URL: https://github.com/bheemisme/time-series-forecasting
• Owner: bheemisme
• Created: 2024-11-29T15:37:22.000Z (over 1 year ago)
• Default Branch: master
• Last Pushed: 2026-04-17T15:19:52.000Z (2 months ago)
• Last Synced: 2026-04-17T17:28:53.151Z (2 months ago)
• Topics: lstm, machine-learning, python, scikit-learn, time-series-forecasting, xgboost
• Language: Jupyter Notebook
• Homepage: https://time-series-forecasting-rzwcldbmtfswfmxufbgswn.streamlit.app
• Size: 5.75 MB
• Stars: 0
• Watchers: 1
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# Time Series Forecasting

![Kaggle Notebook](https://www.kaggle.com/code/sudarshan1927/time-series-forecasting)

- Built a time series forecasting pipeline applying both LSTM (deep learning) and XGBoost (machine learning) models across multiple real-world datasets.

- Worked with diverse datasets, including stock prices (Yahoo, Apple), gold ETF prices, retail sales, and climate temperature data, capturing varied patterns like trend, seasonality, and periodicity.

- Performed comprehensive data preprocessing, including cleaning, handling missing values, and transforming non-stationary series for model readiness.

- Trained and evaluated LSTM and XGBoost models for each dataset, comparing performance across training and test sets to analyse generalisation behaviour.

- Observed that LSTM performed robustly on non-stationary financial data, while XGBoost excelled on stationary retail sales data; both models performed well on periodic temperature data, highlighting dataset-dependent model suitability.