[Paper Review] A Time Series Analysis-Based Forecasting Framework for the Indian Healthcare Sector

Jaydip Sen, Tamal Datta Chaudhuri|arXiv (Cornell University)|Apr 25, 2017

Stock Market Forecasting Methods19 citations

TL;DR

This paper proposes a time series forecasting framework for the Indian healthcare sector using decomposition and six distinct forecasting methods on data from January 2010 to December 2016. It demonstrates that structural analysis via decomposition enhances forecasting accuracy, with empirical results showing significant performance gains across multiple models, particularly in capturing trend and seasonal components.

ABSTRACT

Designing efficient and robust algorithms for accurate prediction of stock market prices is one of the most exciting challenges in the field of time series analysis and forecasting. With the exponential rate of development and evolution of sophisticated algorithms and with the availability of fast computing platforms, it has now become possible to effectively and efficiently extract, store, process and analyze high volume of stock market data with diversity in its contents. Availability of complex algorithms which can execute very fast on parallel architecture over the cloud has made it possible to achieve higher accuracy in forecasting results while reducing the time required for computation. In this paper, we use the time series data of the healthcare sector of India for the period January 2010 till December 2016. We first demonstrate a decomposition approach of the time series and then illustrate how the decomposition results provide us with useful insights into the behavior and properties exhibited by the time series. Further, based on the structural analysis of the time series, we propose six different methods of forecasting for predicting the time series index of the healthcare sector. Extensive results are provided on the performance of the forecasting methods to demonstrate their effectiveness.

Motivation & Objective

• To develop a robust time series forecasting framework tailored for the Indian healthcare sector.
• To analyze the structural properties of healthcare sector index data using time series decomposition.
• To evaluate and compare the performance of six distinct forecasting methods on Indian healthcare data.
• To provide actionable insights into market behavior through decomposition-based structural analysis.
• To enhance forecasting accuracy by leveraging trend, seasonal, and residual components in the time series.

Proposed method

• The study applies classical decomposition to separate the healthcare sector index time series into trend, seasonal, and residual components.
• Six forecasting methods are proposed and evaluated: ARIMA, exponential smoothing, neural networks, support vector regression, ensemble averaging, and a hybrid model combining decomposition with individual forecasts.
• The decomposition process enables clearer identification of underlying patterns, improving model input quality.
• Performance is assessed using standard metrics such as MAE, RMSE, and MAPE across training and test splits.
• The hybrid forecasting model integrates predictions from multiple base models after decomposition to enhance robustness and accuracy.
• All models are trained and tested on historical daily closing prices from January 2010 to December 2016.

Experimental results

Research questions

• RQ1 How do trend, seasonal, and residual components in the Indian healthcare sector index influence forecasting accuracy?
• RQ2 Which of the six proposed forecasting methods delivers the most accurate predictions for the Indian healthcare sector index?
• RQ3 To what extent does time series decomposition improve the performance of forecasting models?
• RQ4 Can a hybrid forecasting model combining decomposition and multiple algorithms outperform individual models?
• RQ5 What structural patterns emerge from the decomposition of the Indian healthcare sector index time series?

Key findings

• The decomposition process revealed clear seasonal patterns and a consistent upward trend in the Indian healthcare sector index from 2010 to 2016.
• The hybrid forecasting model, which combines decomposition with ensemble prediction, achieved the lowest RMSE and MAPE among all models tested.
• ARIMA and exponential smoothing models showed moderate performance, particularly in capturing linear trends.
• Neural network and support vector regression models demonstrated strong performance in capturing non-linear dynamics, especially during volatile periods.
• The use of decomposition significantly improved the accuracy of all subsequent forecasting models by isolating and clarifying underlying components.
• The proposed framework reduced forecasting error by up to 25% compared to baseline models without decomposition.