Master of Business Administration

Faculty / School

School of Business Studies (SBS)


Faisal Jalal

Project Coordinator (External)

Mr. Amad Mughal


Midas Safety Pvt. Ltd

Committee Member 1

Faisal Jalal Visiting Faculty Business Administration - Academic Affairs Institute of Business Administration (IBA), Karachi

Project Type

MBA Research Project

Abstract / Summary

The advancements in modern technology have bred the recent struggle of handling and managing data. 2.5 quintillion bytes of data is generated daily which is expected to increase exponentially with the growing popularity of the internet of things (IoT). This makes the volume of data more significant and relevant with each passing day and this sheer volume of data is what businesses are bombarded with daily. On a global scale, supply chains are faced with five major challenges namely: cost containment, supply chain visibility, risk management, increasing customer demand, and globalization. This sheer load of data does not make it easier to handle either of the major challenges since drawing out inferences for decision-making is becoming more and more of a hassle. It is necessary to bring a modern solution to this modern problem, which is big data analytics.

Big Data analytics is the discovery and communication of meaningful patterns in data through simultaneous application of statistics, computer programming, analytical tools, mathematics, and operation research to quantify and evaluate performance. The use of big data not only equips supply chains with better data accuracy but also improves clarity and insights ultimately driving supply chains forward to a greater e-contextual intelligence shared across the downstream and upstream. Its availability is an asset for businesses that must be utilized to analyze useful variables and make informed decisions.

The textile trade is primarily involved with the planning, production, and distribution of yarn, textile, and covering. Pakistan is the 8th major exporter of textile goods in Asia. It is the 4th largest manufacturer and 3rd largest purchaser of yarn. Textile Industry in Pakistan comprises 46% of the total manufacturing sector and provides employment opportunities to 40% of the total workforce. 5% of the total textile firms are listed on the PSX. The textile sector has a major effect on Pakistan's economic activities, as demonstrated by its direct contribution to domestic development, financial services, and foreign exchange earnings. Despite the textile production being heavily automation-based, it is still termed as a virgin area with regards to Industry 4.0. However, it is expected that efficiency will significantly improve once the fruits of data analytics technology are incorporated into the sector. A dearth of data sharing is observed in the sector when it comes to data mining and machine learning studies due to commercial and confidentiality concerns.

Keeping this in view, we initiated a project alongside Midas Safety, to bring out insights from their raw data and develop a demand forecasting model, using supply chain analytics. This way we attempted to improve their demand forecasting accuracy while benefiting the organization in reducing unnecessary costs associated with extra inventory holding, sell loss, etc.

To build a demand forecasting model for Midas Safety, sales data of the company was acquired. After several semi-structured interviews with their management, an understanding was developed of the patterns in data. We extensively consulted published articles and journals, as part of our secondary research. We worked on data cleaning and identifying the key variables affecting the demand of the company. After the data cleaning, we analyzed the data and developed As-Is-Scenarios for the top 3 customers i.e., Alexandra, Dimensions, and Dickies. For each of these customers, the top categories with significant enough data were chosen to see the purchase trend of each of these customers. We found that under the As-Is-Condition model, there were a lot of variations in the purchasing patterns of the top 3 selected customers. Next, What-If-Scenarios was created to see if the variations in the data can be reduced, and the accuracy of forecasting can be increased. We ran four forecasting techniques i.e., Naïve, Moving Average, Weighted Moving Average, and Exponential Smoothing on the top 3 selected customers. As a result, the accuracy of the forecasting technique improved, which showed a positive sign towards the What-If-Scenario.

A category-wise analysis was also done on the data and where As-Is-Condition for different categories was made to see if the purchase trend can be identified. To see the accuracy and improvement in the forecasting techniques, 4 forecasting techniques were used on all 8 categories in the data. A positive trend was observed as the accuracy of forecasting techniques improved under What-If-Scenario as compared to As-Is-Scenario.

To conclude, we would recommend the organization to use What-If-Scenario models 4 and 5 as it yields better results and increases the forecasting accuracy. However, there is still room for detailed research where huge chunks of data can be used to determine the actual prowess of the models as more data is incorporated in the models, accuracy tends to improve.

Available for download on Sunday, July 18, 2027

The full text of this document is only accessible to authorized users.