Skip to content

Airline passenger traffic prediction using time series forecasting techniques

Notifications You must be signed in to change notification settings

DataScienceVishal/Airline_Passenger_Prediction

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 

Repository files navigation

Project Title: Airline Passenger Traffic Prediction with Trends and Seasonality

Description

The project aims to predict airline passenger traffic using time series forecasting techniques while considering trends and seasonality. It involves analyzing 144 months of monthly airline passenger data. Various time series forecasting methods are implemented and evaluated, including simple methods like naive forecasting, simple average, and simple moving average, as well as exponential smoothing methods like simple exponential smoothing, Holt's method with trend, and Holt Winter's method with trend and seasonality.

Steps Taken

  1. Importing Required Packages: Pandas, NumPy, Matplotlib, and Seaborn were imported for data manipulation, visualization, and analysis.
  2. Importing Time Series Data: Airline passenger traffic data was loaded from a CSV file and formatted to include a datetime index.
  3. Time Series Analysis: Initial analysis of the time series data was conducted, including visualizing the passenger traffic over time.
  4. Missing Value Treatment: Missing values in the dataset were handled using mean imputation and linear interpolation methods.
  5. Outlier Detection: Outliers in the passenger traffic data were identified using box plots and histograms.
  6. Time Series Decomposition: The time series data was decomposed into its trend, seasonality, and residual components using both additive and multiplicative decomposition methods.
  7. Build and Evaluate Time Series Forecast: The dataset was split into training and test sets, and various time series forecasting methods were implemented and evaluated using metrics such as RMSE (Root Mean Squared Error) and MAPE (Mean Absolute Percentage Error).
    • Naive Method: Predicting the next value in the time series using the last observed value.
    • Simple Average Method: Predicting the next value using the average of all past observations.
    • Simple Moving Average Method: Predicting the next value using the average of a fixed window of past observations.
    • Simple Exponential Smoothing: Predicting the next value using a weighted average of past observations, with exponentially decreasing weights.
    • Holt's Exponential Smoothing Method with Trend: Incorporating a linear trend component into exponential smoothing.
    • Holt Winter's Additive Method: Incorporating both trend and seasonality into exponential smoothing.
    • Holt Winter's Multiplicative Method: Similar to the additive method but using multiplicative seasonality instead.
  8. Results and Conclusion: The performance of each forecasting method was compared, and the results were summarized. The Holt Winter's additive method with trend and seasonality achieved the lowest RMSE and MAPE, indicating its effectiveness in predicting airline passenger traffic.

Dataset Information

  • Source: Airline passenger traffic dataset
  • Length: 144 months (12 years)
  • Frequency: Monthly
  • Features: Month (datetime), Passengers (number of airline passengers)

Key Findings

  • Various time series forecasting methods were implemented and evaluated for predicting airline passenger traffic.
  • The Holt Winter's additive method with trend and seasonality achieved the lowest RMSE and MAPE, indicating its effectiveness in predicting passenger traffic.

Future Work

  • Further exploration of advanced time series forecasting models and techniques.
  • Incorporation of additional external factors that may influence airline passenger traffic, such as economic indicators or travel policies.

For detailed implementation and code, refer to the Jupyter notebook provided in this repository.

Feel free to modify and expand upon this description based on your specific project details and preferences.