/Aireline-passengers-time-series-prediction

Primary LanguageJupyter Notebook

Air Passengers Time Series Prediction

Repository Overview

This repository contains two distinct approaches to predicting air passenger numbers using time series forecasting techniques:

  1. Recurrent Neural Network (RNN) with LSTM
  2. Custom ARIMA-like Model

Dataset

  • Source: Air Passengers Monthly Data
  • Description: Monthly total number of international airline passengers from 1949 to 1960
  • Objective: Predict future passenger numbers using different methodologies

Project Structure

air-passengers-prediction/
│
├── data/
│   └── AirPassengers.csv
│
├── rnn_prediction.py
├── arima_prediction.py
│
├── README.md
└── requirements.txt

Approaches

1. RNN (Recurrent Neural Network) Prediction

  • Model: LSTM-based neural network
  • Key Features:
    • Deep learning approach
    • Sequence-to-sequence prediction
    • Captures complex temporal dependencies
  • Preprocessing:
    • Min-Max scaling
    • Sequence creation
    • 3D tensor transformation

2. Custom ARIMA-like Prediction

  • Model: Linear Regression with RNN-inspired preprocessing
  • Key Features:
    • Statistical time series approach
    • Custom sequence preparation
    • Flexible model parameters
  • Preprocessing:
    • Similar to RNN approach
    • Linear regression-based prediction

Dependencies

  • NumPy
  • Pandas
  • Matplotlib
  • Scikit-learn
  • TensorFlow (for RNN)

Installation

  1. Clone the repository
git clone https://github.com/yourusername/air-passengers-prediction.git
cd air-passengers-prediction
  1. Install dependencies
pip install -r requirements.txt

Usage

RNN Prediction

python rnn_prediction.py

ARIMA-like Prediction

python arima_prediction.py

Performance Metrics

Both models are evaluated using:

  • Mean Squared Error (MSE)
  • Mean Absolute Error (MAE)

Visualization

Each implementation generates:

  • Actual vs Predicted Passenger Numbers
  • Loss Curves (for RNN)
  • Prediction Error Analysis

Comparative Analysis

RNN Strengths

  • Captures non-linear patterns
  • Learns complex temporal dependencies
  • Deep learning approach

ARIMA-like Model Strengths

  • Simpler implementation
  • Statistical foundation
  • Faster computation

Limitations

  • RNN:

    • Requires more data
    • Computationally expensive
    • Hyperparameter sensitive

  • ARIMA-like Model: