5380
Decision Support System for Small Enterprises to Revive themselves in the Covid Era
using Lessons of 2008-09 Recession
Prashant Kumar Mishra
a, Soumyadeep Bhattacharya
b, S.S. Sridhar
caB. Tech, Computer Science and Engineering, SRM Institute of Science and Technology, Kattankulathur, Tamil Nadu, India. E-mail:
ps4629@srmist.edu.in
bB. Tech - Computer Science and Engineering, SRM Institute of Science and Technology, Kattankulathur, Tamil Nadu, India. E-mail: sg8171@srmist.edu.in
cProfessor, Computer Science and Engineering, SRM Institute of Science and Technology, Kattankulathur, Tamil Nadu, India. E-mail: sridhars@srmist.edu.in
Article History: Received: 11 January 2021; Revised: 12 February 2021; Accepted: 27 March 2021; Published online: 10 May 2021
Abstract: The current global economic scenario is disturbing and needs immediate attention and innovative solutions. The Corona Virus disease 2019 (Covid-19) has already claimed more than 2.5 million lives and has pushed millions of people into poverty and all the nations in an economic crisis. Small enterprises are no exception and have been severely affected. This economic situation is termed by leaders and politicians across the globe as “new” or “unforeseen” but a closer look reveals that it is not that new to us and the previously employed methods can work effectively. This paper is an attempt provide solutions to the small enterprises which were employed during the recession of 2008-09. This paper starts by surveying various shops and establishments, collecting relevant data of both today and 2008- 09. This paper then develops a Decision Support System which takes into account the enterprises’ data and provides steps taken in the past that can yield good returns. The proposed system will prove to be very useful considering the current situation and can revive small enterprises, and thus, economies.
Keywords: Corona Virus, Covid-19, Recession of 2008-09, Decision Support System, Decision Tree. DOI:10.16949/turkbilmat.702540
1. Introduction
The Coronavirsus disease 2019, better known as the Covid- 19 is a contagious disease caused by the novel coronavirus. The symptoms are regular like cold, cough, fever, etc. In extreme cases, symptoms like pneumonia, breathing difficulties, loss of smell may also occur. A lot of research is going on currently on this and thankfully, a great deal of results can be seen as well. Various vaccines in record times have been created and people across the world are getting vaccinated. But we still have a long way to go. The way computer science and its disciplines and applications have enabled researchers is remarkable. Data Science, Analytics, Machine Learning and Artificial Intelligence is paving way for remarkable healthcare solutions to deal with the pandemic.
Having claimed more than 2.5 million lives and infecting more than 124 million people, the virus is still spreading with different variants of it all around the globe. Because of its contagious nature, measures across the planet were taken to stop the spread of the disease. These measures included strict lockdowns and curfews, travel restrictions, etc. Naturally, this led to shut down of businesses for a long time. This affected business. The most affected businesses were the small brick and mortar retail establishments, manufacturing units, etc.
Many people find this event as ’new’ or ’unforeseen’, but we believe it is similar to the recession of 2008-09. Both have similar factors like market unrest, low demand, etc. The responses collected from various shops across different cities verify the same.
In this paper, we demonstrate how a decision support system can be formed using the data collected and can be trained to recommend steps for small enterprises to carry out in order to help them.
2. Literature Survey
Although this project is not an extension or further work of any previously done work, it draws its inspiration from some of the work going on currently. We found many papers related to ways in which things can be revived and sustained during these challenging times. A lot of work is going on currently on what ways and strategies to implement during this pandemic and our work will be one of those works.
5381
LJ Muhammad et al in [6] aim to predict the recovery time of infected patients. It provides the minimum and maximum days that it would take for a patient to recover by dividing people in age groups of 3 categories, namely: Patients who are at high risk and are not expected to recover from the virus, Patients who are likely to recover and Patients who are likely to recover quickly. Decision tree, support vector machine, naive Bayes, logistic regression, random forest, and K-nearest neighbour algorithms were all applied and Decision Tree made the most accurate predictions.Tsan-Ming Choi in [1] attempt to compare static operation services with mobile operation services by converting a static business into a mobile business and find if people adapt to it or not.
Afees A Salisu and Xuan Vinh Vo in [10] aim to understand the effect that the health-related news during this pandemic has had on the values of stock returns. Data for the study was taken from the 20 countries that had been worst hit by this pandemic in terms of deaths and cases reported. It was observed from the following study that as health news worsened, that is if there was an increase in the reported cases and deaths, the prices of listings on the stock market also massively dropped. On the other hand, in case on any day the reported deaths and cases were lesser than the previous day then the prices of listings on the stock market increased. An advantage of this model was that it gave more accurate results when compared to traditional models.
Rajiv Chowdhury et al in [2] discuss the impact of lock-downs, etc. in curbing the rate of Covid-19 in 16 countries and tried to bring about innovative solutions to deal with them.
Justus Kithiia et al in [5] provide primary data on social and economical impact of Covid-19 in a city in Kenya. Data was primarily conducted using online surveys and questionnaires. Quoc-Viet Pham et al in [7] emphasize the importance of AI and Big Data analytical models in responding to the Covid- 19 outbreak and how they can be used to prevent the severe effects of Covid-19.
Adedoyin Ahmed Hussain et al in [4] summarize the current state of AI applications in clinical administrations and healthcare industry while battling Covid-19.
Shohini Roy in [8] analyses the economic impact of COVID-19 on the following sectors, namely, tourism industry, oil industry, aviation industry, financial sector and healthcare sector.
All the papers surveyed provided a clear picture of how Covid-19 has affected the world in so many negative ways. The papers also provided an insight into the computational models and algorithms that can be implemented for a better future. The papers also give us an insight into how the effect and spread can be controlled using computational models and algorithms. The papers surveyed have laid a solid ground for our work and has provided us with very useful insights to implement the project’s vision.
3. Proposed Work
The work proposed in this paper can be summarised in the UML diagram given in figure 1 which basically outlines the data provided by the user and the output provided by the system after working on the data.
5382
Figure 1. UML Diagram of the proposed system
The implementation of the above working model can be understood by the architecture diagram given in figure 2.
A. Abbreviations and Acronyms
The following acronyms and abbreviations will be used: • DSS: Decision Support System
• CSV: Comma-separated values • INR: Indian Rupee
4. Implementation
The DSS is developed in Python in the Google Colab environment.
The implementation can be understood by understanding the three modules as shown in the architecture diagram.
5383
Figure 2. Architecture Diagram of the proposed system. A. Module 1: Data collection and Processing
The required data is collecting via surveying various shops and filling out the answers in the Google form created for the purpose. The main questions that were used to create the required dataset for the DSS included
• Type of business (Manufacturing/Retail/Wholesale/Ser- vices)
• Industry (Building materials/Textile/Toy/IT/Food/Station- ary/...)
• Annual Revenue (in INR) before Covid-19
• Revenue earned as a percentage over last year’s revenue (as provided in the above question) • Steps taken during the recession of 2008-09.
The data is then downloaded from the Google form in a CSV file. This CSV file is then moved to Google Drive for easy sharing between the participants. The file is then loaded into Google Colab and transformed into a working dataset. Any missing data is then filled accordingly.
Some visualizations of our data to understand the nature of data are given in figures 3, 4, 5 and 6. B. Module 2: Generating Quick Insights about the Data
A new dataset is obtained from the original dataset in the first module. This new dataset serves as the data for our DSS. Using general functions provided by Python libraries, some quick insights about the data are tabulated in Table 1:
C. Module 3: Designing the Decision Support System
To design our DSS, we used Decision Tree for better results. We start by dividing our new dataset derived in module 2 in X and Y. X contains 4 columns, which are the 4 features.
5384
Figure 3. On a scale of 1 to 5, how much has the business been affected by Covid-19?
Figure 4. Reasons attributed to losses due to Covid-19.
Y contains just 1 column which is our target value. Since our features in X contain categorical values, we have encoded them to integers with the help of label encoder function. Dataset Y is factorized due to the same reason with the help of factorize function.
The datasets are then split into training and testing data with 75% as training and the remaining 25% as test. Table I. Quick Insights
Description Result
Most stated reasons for losses dur- ing Covid-19
Shut down of firm due to lockdown. Low demand by customers. Supply Chain Disruptions. Market unrest.
Low inventory. On a scale of 1 to 5, 1 being least
severe and 5 being most severe, most firms rated the losses due to Covid-19 as
5 Based on the responses received,
most firms find the events of Covid-19 and recession of 2008-09
Similar Most stated reasons for losses dur- ing the recession of
2008-09
Low demand by customers. Market unrest.
5385
Figure 6. Reasons attributed to losses due to the recession of 2008-09
5386
The accuracy of our DSS came out to be 0.88. The accuracy was measured by the metrics.accuracy score provided by sklearn library.The user is prompted to enter their details as visualized in the UML diagram in Figure 1. The data is then stored in the test dataset.
The training and testing datasets are then taken into account by the Decision Tree Classifier which trains itself. It then pre- dicts the test dataset which also contains information entered by the user. The output is an integer which corresponds to a value in the steps.
Because of its very large nature, a part of the decision tree after its implementation is visualized in figure 7.
5. Results
The DSS after taking into account the user’s data provides steps of the past taken by a similar firm as the output.
The output of the DSS for a particular case is given below:
Thanks for your information. Based on your input,
we find that a firm similar to you followed the following steps back during the recession of 2008-09:
Financial support from banks
and other financial institutions.; Reduced inventory.; Sold assets.
6. Conclusion
The proposed system aims to help owners and managers of small establishments to revive their business in these tough times. The proposed system wishes to provide owners and managers with solutions to cope up with the challenges faced because of Covid-19. The project’s scope reaches out from sole proprietors based retail shops to small scale enterprises, both manufacturing and retail, and to the services establishments.
We hope this research is used by the administration of various establishments and helps them in improving their profits. We hope to find more data to improve our model to further help as much as possible.
Future prospects based on this project may include location and other important factors also as the features. We hope that this research becomes the base for more research and work in similar fields to help businesses and economies to revive.
Acknowledgement
We are thankful to the administration and staff at the SRM Institute of Science and Technology, Kattankulathur for their support, suggestions and belief throughout our research.
References
1.
Tsan-Ming Choi. “Innovative “bring-service-near-your-home” operations under Corona-virus (COVID- 19/SARS-CoV-2) outbreak: Can logistics become the messiah?” In: Transportation ResearchPart E: Logistics and Transportation Review 140 (2020),
p. 101961.
2. Rajiv Chowdhury et al. “Dynamic interventions to control COVID-19 pandemic: a multivariate prediction modelling study comparing 16 worldwide countries”. In: European journal of epidemiology 35.5 (2020), pp. 389–399.
3. Kannan Govindan, Hassan Mina, and Behrouz Alavi. “A decision support system for demand management in healthcare supply chains considering the epi- demic outbreaks: A case study of coronavirus disease 2019 (COVID-19)”. In: Transportation Research Part E: Logistics and
Transportation Review 138 (2020), p. 101967.
4. Adedoyin Ahmed Hussain et al. “AI techniques for COVID-19”. In: IEEE Access 8 (2020), pp. 128776–
