Elsevier

Pattern Recognition Letters

Volume 152, December 2021, Pages 172-179
Pattern Recognition Letters

Pattern Recognition Letters
Prognosticating the effect on Unemployment rate in the post-pandemic India via Time-Series Forecasting and Least Squares Approximation

https://doi.org/10.1016/j.patrec.2021.10.012Get rights and content

Research Highlights (Required)

  • A novel time-series forecasting architecture that helps deal with sudden hike in the data.

  • A pattern recognition architecture that helps to forecast unemployment rate in India in the post-pandemic timeframe.

  • The current study proposes a novel architecture to deal with the rare unusual trends that are not seasonal in nature.

  • Vanilla time-series forecasting frameworks remained ineligible to capture the rare sudden patterns.

Abstract

The current paper aims to analytically visualize the future outcomes that the post-pandemic India might have in store for its citizens. We use time series forecasting on various collected data and combined the statistics of economics-deciding parameters to forecast the trends that might be prevalent in the next year. Since, the data contains a single anomalous trend, even the Prophet model could not learn this property from the data since this trend is not seasonal in nature. The current study proposes a novel architecture to deal with these rare unusual trends by combining two models - one learning normal usual patterns and the other getting trained on usual as well as rare anomalous patterns. It could help in dealing with sudden hike patterns like due to COVID-19 in the data, and lead to better forecasting on future timeframes. We combined the results of two distinct time-forecasting models trained on two sets of data of varying timeline lengths, using parameters obtained from Least Squares Approximation (LSA). The LSA helps us find an approximate vector approximation so as to obtain a model performing closely to the actual.

Introduction

COVID-19 pandemic has hit the world very harshly. India has now become a hotspot of this pandemic after the USA and Brazil, after surpassing Russia in the number of cases, and is soon to enter stage 3 considering fatality rates. The countries are helpless before this natural phenomenon and are forced to implement lockdowns to prevent its spread. As is evident from many other historical pandemics and epidemics in various parts of the world, there has always been a great deal of adverse effects on the economic conditions of the country or the world as a whole. Consequently, due to lockdowns, trading and working in the industries is stalled which in turn leads to loss of jobs of numerous employees. The employers are then not in the position to support themselves and thereby many companies and industries get shut down. Since industries run a country's economy, and with the industries shutting down or getting in loss, the economy of the entire country gets disrupted. A disease outbreak, being at the scale of the world, affects the entire world. Since most countries are connected through trade and commerce, the pandemic can lead to utter loss of jobs and economy of nations.

The COVID-19 disease has been growing steadily at an exponential rate throughout the world. Since its inception, it has disrupted various sectors of human life including business, health, education and governance. As of July 20, 2020 the number of cases is 14,360,451, out of which 603,285 have lost their life while 8,071,937 have recovered1. The economies have been adversely affected and their GDP (Gross Domestic Product) is following a negative slope in today's scenarios. In India, due to sequence of lockdowns, menial laborers have lost their daily-wage jobs, and are left with nothing to sustain in this pandemic-affected economy, and hence many have died not due to the disease but of scarcity of money to support one's family. This is a very arduous situation in the country, and its citizens must not leave hope. Government has been doing its best to control this pandemic. It has taken many initiatives and cut down the cost of COVID-19 tests by half, but still the situation in the country is not under control and is surging high.

The unemployment rate in India had been swinging closely to around 2.7% in previous years but has been exaggerated to an astonishing 15% average during this pandemic, especially between March and May. The rules of social distancing and lockdown has added more to disruption rate of dwindling unemployment rate and consequently on the overall economy of the country. Apart from studying the effects of the pandemic on unemployment rate, inflation rate and economy growth rate, the paper also analyses the consequences of up-skilling at home, on academia and corporate world. We run time-series forecasting on the collected data, and present our findings through comprehensive analysis of the results.

The paper is organized as follows. Section 2 presents a discussion about the works that have been conducted and published in the past. Section 3 discusses the algorithms and time-series forecasting architectures used in the paper. Section 4 outlines the methodology that has been adopted to implement analyses presented in the paper, and also describes the workflow that is used in the implementation of the proposed architecture. Section 5 talks about the issue of data unavailability, its collection and curation. Section 6 presents the results of the algorithms employed to perform time-series forecasting on the curated dataset. Section 7 presents a comparative analysis of the proposed architecture with other available time-series forecasting frameworks and existing researches. The several engineering applications of time-series forecasting is presented in Section 8. At last, Section 9 summarizes and concludes the paper.

Section snippets

Related work

With the onset of the pandemic, a lot of researches have been conducted by researchers worldwide on the various aspects of the mankind's life from health and hygiene to social and environmental contexts. There exists the study where the author estimated the date when the number of cases would reach the peak and when it will flatten out using methods shared from data analysis and machine learning paradigm [1]. The authors in [2] used the time-series forecasting methods like ARIMA to predict the

Least Squares Approximation

LSA is a method to approximate the solution to a set of linear equations for whom a real solution does not exist. It works by minimizing the sum of squares of residuals obtained by each single equation. It is most commonly used in the domain of data fitting, wherein the residuals are the difference between observed and actual values, and the LSA tries to minimize the sum of these residuals.

Since, a line, representing solution, intersecting with the plane of set of equations, cannot be found;

Heuristic-based Model Training and Prediction

The prediction of future effects has been made using Facebook's time-series forecasting framework Prophet. The accumulated data was first cleaned and pre-processed as per the requirements. Due to the pandemic in 2020 and unavailability of required data, basic implementations of these models could not perform and give reliable results.

To overcome the predilection of the learning models on comparatively larger pre-pandemic historical data and the pandemic-induced sudden hike in unemployment rate,

Data Collection

There is an unavailability of proper data regarding unemployment rate in India prior to 2016. Adequate data started to be gauged from 2016 onward by CMIE3, India. But this much amount of data is not adequate and sufficient for training statistical and machine learning models - they need good amount of data to learn patterns. So, we also referred to World Bank and the International Labor Organization (ILO) for the data. After data aggregation and cleaning, the data before

Results and Discussion

The experimentation and inferencing involves three steps. In the first step, we run the time-series forecasting framework on two types of datasets that we have created. Thereafter, we model the results in the form of matrix equations, and then in the last step, we combine the results and optimize it by finding suitable parameters a andb using Least Squares Approximation (LSA). The LSA utilises the Levenberg-Marquardt Algorithm (LMA) [14] underneath to find a vector approximation. All the

Advantages over Other Vanilla Models

As shown in Section 6, the vanilla models (except Prophet) could not correctly learn the pattern in the data. They were either over-hyped due to the sudden COVID-19 hike or they were just not being affected by this anomalous trend in the data. Likewise, ARIMA, FARIMA, LSTM and SVR could not learn the pattern, while SARIMA did try to learn and adjust to the pattern in the data, it still was unable to address the non-stationarity property that is present in the data, it just predicted strictly

Engineering Applications of Time-series Forecasting

Time series forecasting has already proven its importance in various fields like weather forecasting, process and quantity control, stock market predictions etc. The new emerging applications of time series analysis can be found in energy consumption sectors. One of such works is reported in [17], in which the new building's energy consumption where the past data does not exist, computer simulation is used in such scenarios to predict future energy consumption. While for existing buildings,

Conclusion

The COVID-19 pandemic adversely influenced and disturbed all the phases of human lives. The number of cases in India has crossed an enormous point of 1.44 millions surpassing Russia and is now third on the list, just behind USA and Brazil. An alarming situation has engulfed whole of the country with many places identified as Containment and Red zones. The search for the cure for this disease is surging high but in due time countries are helpless. They are forced to bow before this pandemic and

Authorship Confirmation

Please save a copy of this file, complete and upload as the “Confirmation of Authorship” file.

As corresponding author I, Pawan Singh, hereby confirm on behalf of all authors that:

  • 1.

    This manuscript, or a large part of it, has not been published, was not, and is not being submitted to any other journal.

  • 2.

    If presented at or submitted to or published at a conference(s), the conference(s) is (are) identified and substantial justification for re-publication is presented below. A copy of conference

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

References (19)

  • C. Deb et al.

    A review on time series forecasting techniques for building energy consumption

    Renewable and Sustainable Energy Reviews

    (2017)
  • S. Sengupta et al.

    Covid-19 pandemic data analysis and forecasting using machine learning algorithms

    (2020)
  • R. Gupta et al.

    Trend analysis and forecasting of covid-19 outbreak in india

    (2020)
  • A.K. Poddar

    Impact of covid-19 on indian economy-a review ajay kumar poddar and brijendra singh yadav2

    Horizon

    (2020)
  • S.M. Dev et al.

    Covid-19: Impact on the indian economy

    (2020)
  • D.Kumar Das et al.

    The impact of covid-19 in Indian economy–an empirical study

    International Journal of Electrical Engineering and Technology

    (2020)
  • C. Katris

    Prediction of unemployment rates with time series and machine learning techniques

    Computational Economics

    (2020)
  • M. Desaling Germay, Modeling and forecasting unemployment rate in sweden using various econometric measures...
  • M. Fajar, O. R. Prasetyo, S. Nonalisa, W. Wahyudi, Forecasting unemployment rate in the time of covid-19 pandemic using...
There are more references available in the full text version of this article.

Cited by (3)

2012 Elsevier Ltd. All rights reserved.

View full text