Time Series Modelling and Forecasting using R.

On 17th September 2025, Christ College, Pune organized an insightful guest lecture on “Time Series Modelling and Forecasting using R.” The session was conducted by Ms. Divya K. Andrew, Senior Research Fellow at Cochin University of Science and Technology, Kerala. The one-hour session, held from 10:30 AM to 11:30 AM, was attended by final year MSc. Data science students. The class was organized under the guidance of Ms. Nisha Mary Daniel, the course instructor for Time Series at Christ College, Pune.

The objective of the session was to provide participants with an in-depth understanding of time series concepts and guide them in constructing forecasting models using the R programming language. Ms. Divya began with an overview of the importance of time series data, emphasizing its applications in financial markets, demand forecasting, economic planning, weather prediction, and operations management.

The lecture started with a discussion on the components of a time series-trend, seasonality, cyclic variations, and random noise. Ms. Divya highlighted that identifying these components is crucial for accurate forecasting. She explained the concept of stationarity, demonstrated how to check for it using plots and tests, and stressed its importance in building valid models.

The core section of the lecture focused on different types of time series models and their construction in R:

• Autoregressive (AR) Model: Ms. Divya explained how AR models rely on the dependency of current observations on their past values. She demonstrated the use of lag plots and R functions to estimate AR parameters.
• Moving Average (MA) Model: The discussion moved to MA models, where predictions are based on past error terms. She distinguished between AR and MA models, highlighting when each is applicable.
• Autoregressive Moving Average (ARMA) Model: As a combination of AR and MA, the ARMA model was introduced as an effective tool for modeling stationary series. Ms. Divya presented an R-based example to illustrate its implementation and evaluation.
• Autoregressive Integrated Moving Average (ARIMA) Model: Finally, Ms. Divya explained how ARIMA extends ARMA by incorporating differencing to handle non-stationary data. She elaborated on the meaning of the parameters (p, d, q) and guided participants on how to identify suitable orders using autocorrelation (ACF) and partial autocorrelation (PACF) plots.
  

A brief hands-on demonstration followed, where Ms. Divya showcased the use of popular R packages like forecast and tseries for building models, estimating parameters, and generating forecasts. She also underlined the importance of splitting datasets into training and testing sets and evaluating models using metrics such as Mean Absolute Error (MAE) and Root Mean Square Error (RMSE).

The session concluded with an interactive Q&A, where participants clarified doubts related to model and interpretation of results. Ms. Divya encouraged students to experiment with real datasets, stressing that practical application would strengthen their understanding of forecasting techniques.

Overall, the session was highly enriching and interactive, blending theory with practical demonstrations. It provided participants with valuable insights into time series modeling and motivated them to apply these methods in academic projects and professional research. The Department of Data Science expressed gratitude to Ms. Divya K. Andrews for delivering such a clear and impactful lecture.