A practical time series forecasting project applying both traditional Machine Learning techniques and advanced Deep Learning architectures. The repository includes full data preprocessing, visualization, and prediction workflows on real-world time series datasets: avocado prices and vehicle miles traveled.
👉 Forecasting avocado prices using:
Linear Regression
Random Forest Regressor
Prophet (Facebook’s time series model)
📌 Includes:
Data loading & sorting by date.
Data exploration (info(), describe()).
Handling time series ordering.
Applying and evaluating ML models.
Model comparison using MAE / MSE.
👉 Forecasting vehicle miles traveled using:
LSTM Neural Network (Keras Sequential model)
📌 Includes:
Data exploration & visualization (plot()).
Setting time-based index.
Feature scaling using MinMaxScaler.
Data reshaping for LSTM input.
Building and training an LSTM model.
Forecasting and plotting predictions.
Evaluating performance via MSE / MAE.
📊 Tools & Libraries: Python
Pandas, Numpy
Matplotlib, Seaborn
Scikit-learn
Keras (TensorFlow backend)
Prophet
Real-world time series datasets.
End-to-end workflow from data preprocessing to model training and prediction.
Clear visualizations to interpret trends and model performance.
Hybrid use of classical ML regressors and sequence-based DL models.
Hands-on implementation of Prophet and LSTM for time series.