Assignment - II

TOPIC: The need for demand forecasting

RESEARCH
METHODOLOGY

SEMESTER VII

Submitted to: Dr. Abhijeet Mukherjee

Submitted By:
Mehul Chandra (BFT/20/198)
Gairika Saha (BFT/20/346)
ABSTRACT:

This study delves into the pivotal realm of demand forecasting within the apparel industry, aiming to
unravel the importance of this practice for consumers. Through a global survey capturing diverse
demographics, the research scrutinizes general trends in consumer demand during holidays and
weekends. The insights garnered are indispensable for a forthcoming graduation project focusing on
supply chain demand forecasting, wherein the XGBoost algorithm will be employed for heightened
predictive accuracy. By exploring consumer preferences and behaviors, this study lays the
groundwork for the development of advanced predictive models in supply chain management, crucial
for optimizing inventory and meeting consumer needs effectively.

Keywords: Demand Forecasting, Apparel Industry, Consumer Preferences, Supply Chain Management,
XGBoost Algorithm, Predictive Models, Inventory Optimization, Consumer Behaviour.

LITERATURE REVIEW:

One of the research papers investigates the feasibility and comparative analysis of Deep Learning
approaches to forecasting the demand problem with implementation to a public data-set. The
authors highlight that time-series prediction has stirred up wide attention in many research fields
because it is an important direction of dynamic data analysis and processing. The prediction is used
in a variety of practical situations, especially for the pressing demand for forecasting future data
trends based on historical information.[1]

While demand forecasting remains a challenge, developments in machine-learning have provided

dramatic improvements. The authors use comparison with RMSE performance metrics to analyze the
Deep Learning performance better than other model techniques, including Random Forest, Gradient
Boosted Trees, and Support Vector Machine. [1]

However, the forecasting problem is a vital need for business decision making. The more insight into
demand from product design to production, logistics to sales, the better prepared the organization
will be. On the contrary, if the forecasts were unsatisfied, it will result in the risk of over or under
production, weakness in service, or a mere selling of the faulty commodity. [1]

Another research paper explores the use of machine learning algorithms for demand forecasting.
The authors adopted the extreme learning machine and other machine learning algorithms such as
K-nearest neighbors, decision tree, gradient boosting, and multi-layer perceptron to examine their
accuracy and performance. The findings demonstrated that artificial neural network-based methods
outperformed the other employed techniques. Furthermore, the extreme learning machine was
found to be the best or at least among the best predictors. [2]

Another research paper compares traditional statistical and machine learning algorithms for demand
forecasting of 50 products. The author uses ARIMA, Theta method, MLP, and Driver-less AI for
forecasting. The results show that ARIMA accurately forecasts the demand for 10 products, Theta for
25 products, and MLP for the remaining 15 products. [3]

The Driver-less AI was outperformed by ARIMA, Theta, and MLP for most of the products. The
research concludes that different algorithms should be implemented when forecasting the demand
of a set of products to identify the best performing algorithm for each product. [3]

The research paper titled "Demand Forecasting Using Neural Network for Supply Chain Management"
investigates the use of Artificial Neural Network (ANN) algorithms for demand forecasting. The authors
highlight that ANN algorithms are useful for demand forecasting due to their ability to accommodate
non-linear data and capture subtle functional relationships among empirical data, even when the
underlying relationships are unknown or hard to describe. [4]
The demand analysis for a valve manufacturing industry, which typically represents a make-to-order
industry, has been carried out using a neural network based on different training methods. The
accuracy of each algorithm is evaluated using the Back-testing technique by splitting the existing data
into a train and test set. The results show that ANN algorithms perform well in demand forecasting,
providing valuable insights for supply chain management. [4]

Another research paper investigates the use of Deep Learning methods, including Auto-Regressive
Integrated Moving Average (ARIMA) and Long Short-Term Memory (LSTM), for demand forecasting in
the context of Supply Chain Management 4.0. The authors highlight that demand forecasting plays a
crucial role in maintaining the balance between demand and supply, thus improving decision making.
A large amount of data is generated throughout the Supply Chain (SC), and Artificial Intelligence (AI)
can consume this data to allow each actor in the SC to gain in performance but also to better know and
understand the customer. The study aims to improve the performance of the demand forecasting
system of the SC based on Deep Learning methods. The experimental results enable the selection of
the most efficient method that could provide better accuracy than the tested methods. [5]

The research paper titled "Data Analytics in the Supply Chain Management: Review of Machine Learning
Applications in Demand Forecasting" provides a comprehensive overview of machine learning
applications in demand forecasting. The authors retrieved a total of 1870 papers from Scopus and Web
of Science databases related to machine learning. Out of these, 79 papers focusing on demand
forecasting were comprehensively reviewed and used for the analysis in this study.[6]

The results showed that neural networks, artificial neural networks, support vector regression, and
support vector machine were among the most widely used algorithms in demand forecasting with 27%,
22%, 18%, and 10%, respectively. This accounted for 77% of the total reviewed articles. Most of the
machine learning application (65%) was applied in the industry sector, and a limited number of articles
(5%) discussed the agriculture sector. The practical implication of this paper is in exposing the current
machine learning issues in the industry to help stakeholders and decision-makers better plan
transformation actions. [6]

The result of on of the papers was to evaluate and compare the different models fairly, the choice of
evaluation metrics was important as each metric has different characteristics. It was important to
include several metrics since different metrics could display different flaws or benefits in the models.
Root mean squared error (RMSE), mean absolute error (MAE), and Mean Absolute Percentage Error
(MAPE) have been used extensively in the academic literature and could, be deemed to be the most
useful. [7]

In comparison with these, when analysing the online communities, data science competitions, and
sources outside the academic literature, it was clear that Symmetric Mean Absolute Percentage Error
(SMAPE) can be beneficial when comparing the models. Parameter Tuning is an important part that is
usually done to achieve good accuracy, fast result, and to deal with over-fitting. [7]

Demand forecasting is one of the main problems of supply chains to optimize stocks, reduce costs, and
increase sales, profit, and customer loyalty. To overcome this issue, there are several methods such as
time series analysis and machine learning approaches to analyze and learn complex interactions and
patterns from historical data. There is a novel attempt to integrate the 11 different forecasting models
that include time series algorithms, support vector regression model, and deep learning method for
demand forecasting process. [8]

The other novelty of this work is the adaptation of boosting ensemble strategy to the demand
forecasting model. In this way, the final decision of the proposed system is based on the best algorithms
of the week by gaining more weight. [8]

Their another objective is to use heuristic methods MBO (Migrating Birds Optimization) and other
related algorithms [50] to optimize some of coefficients/weights which were determined empirically by
trial-and-error like taking 30% percent of the best performing methods in their current system. [8]
METHODOLOGY:

In this review paper, a survey methodology was employed to gather insights into consumer preferences
and behaviors related to apparel product purchasing. The survey targeted individuals of diverse
demographics, including gender, age groups, and employment status, with the aim of obtaining a
comprehensive understanding of the factors influencing their apparel purchasing decisions. The
questionnaire covered various aspects such as frequency of apparel product purchases, the significance
of product availability, and the impact of holidays or weekends on buying habits. Participants were also
asked to rate the importance of demand forecasting and identify factors influencing their purchasing
decisions, such as price, quality, brand, style, comfort, and availability. The survey was distributed
electronically, and responses were collected anonymously to ensure candid and unbiased feedback.
The inclusion of participants from different countries further enriched the data, providing a global
perspective on consumer preferences in the apparel industry. The gathered information serves as a
valuable foundation for the comprehensive review and analysis presented in this paper.

INFERENCE DRAFTED FROM THE GOOGLE FORMS:

Inference on Gender and Demand Forecasting:

Analyzing the gender distribution of survey respondents is crucial for demand forecasting in
the apparel industry. In this survey, the majority of respondents identify as female (67), while
34 identify as male. Understanding these demographic details is vital for accurate demand
forecasting, as it allows businesses to tailor their inventory and marketing strategies based on
the predominant gender group.
Females, being the majority in this survey, may influence the demand for specific apparel
products. Demand forecasting models can leverage this information to predict trends and
preferences, ensuring that the inventory is stocked with products that align with the
anticipated demand. Businesses need to consider the unique preferences and buying
behaviors associated with different gender groups, as this can significantly impact the success
of sales and marketing initiatives.
Inference on Age Group and Demand Forecasting:

The dominance of the 18-24 age group among survey respondents, with 75 individuals, is
particularly noteworthy in the context of the survey being distributed in colleges. This age
group's significant representation aligns with the typical demographic of college students.
Understanding this demographic composition is crucial for demand forecasting in the apparel
industry, especially considering the unique preferences and buying behaviors associated with
college-aged individuals.
College students often exhibit distinct trends in fashion and shopping habits influenced by
factors such as budget constraints, current fashion trends, and the desire for comfort and style.
The prevalence of the 18-24 age group suggests that demand forecasting models should focus
on anticipating and meeting the specific needs of this demographic.

Inference on Employment Status and Demand Forecasting:

The diverse employment status of survey respondents provides valuable insights for demand
forecasting in the apparel industry. The majority of respondents identify as students (73),
followed by those employed full-time (15), employed part-time (9), self-employed (3),
unemployed (4), and retired (1).
Understanding the employment status of individuals is crucial for predicting and meeting the
demand for different types of apparel products. Students, being the largest group in this
survey, often have distinct preferences driven by factors like trends, budget considerations,
and the need for versatile clothing. Demand forecasting models should focus on anticipating
and accommodating the unique needs of this significant demographic.

Considering the smaller representation of retired and unemployed individuals, demand

forecasting models may need to account for potential variations in purchasing behaviors
within these groups.
Inference on Frequency of Apparel Purchases and Demand Forecasting:

The frequency of apparel purchases reported by survey respondents provides crucial insights
for demand forecasting in the apparel industry. The distribution indicates varying shopping
habits among participants.
⚫ The majority of respondents make monthly apparel purchases (60 individuals),
suggesting a steady and consistent demand for clothing on a regular basis.
⚫ Weekly purchasers (8 individuals) and yearly purchasers (18 individuals) represent
segments with distinct buying patterns, potentially influenced by factors such as seasonal
trends, promotions, or special occasions.
⚫ Rarely purchasers (19 individuals) and those who never make apparel purchases (2
individuals) form segments with lower frequency, indicating a need for forecasting
models to account for intermittent or negligible demand in certain consumer segments.
⚫ Understanding these patterns is vital for businesses to optimize inventory management
and marketing strategies. For example, monthly purchasers may benefit from loyalty
programs or subscription services, while yearly purchasers may be targeted with seasonal
promotions. Strategies for infrequent purchasers could involve special promotions or
initiatives to increase engagement.

Inference on Product Availability Importance and Demand Forecasting:

Low Importance (Scale 1-4):

⚫ A combined total of 16 respondents (10 at scale 1 and 6 at scale 3) indicated a relatively

lower importance of product availability when purchasing apparel products.
⚫ Individuals in this group may prioritize other factors, such as price, style, or brand, over
immediate product availability.
⚫ For demand forecasting, businesses may need to consider that some consumers in this
segment may be willing to wait for specific items, allowing for more flexible inventory
management.

Moderate Importance (Scale 5-7):

⚫ A combined total of 29 respondents (5 at scale 5, 8 at scale 6, and 16 at scale 7) indicated

a moderate importance of product availability.
⚫ Consumers in this group likely consider product availability as a significant factor but may
be willing to wait or explore alternatives if the desired product is temporarily
unavailable.
⚫ For demand forecasting, businesses should ensure a reasonable level of inventory to
meet the needs of this moderately sensitive group while also considering factors like
marketing and promotions to mitigate potential delays.

High Importance (Scale 8-10):

⚫ A combined total of 58 respondents (18 at scale 8, 14 at scale 9, and 26 at scale 10)

emphasized the high importance of product availability.
⚫ This group places a premium on immediate access to desired apparel products and may
be less patient with delays.
⚫ Businesses need to prioritize maintaining robust inventory levels, efficient supply chains,
and accurate demand forecasting models to meet the expectations of this highly sensitive
group.

Inference on Apparel Purchases During Holidays or Weekends:

Increased Purchases During Holidays or Weekends (More):

⚫ A total of 48 respondents reported an inclination to purchase more apparel products

during holidays or weekends.
⚫ This group represents a potential surge in demand during specific time frames, suggesting
businesses should anticipate and prepare for increased sales during holidays or
weekends.
⚫ Demand forecasting models should account for seasonality and factor in heightened
demand during these periods to ensure adequate inventory levels.
Decreased Purchases During Holidays or Weekends (Less):

⚫ A smaller segment of 10 respondents reported a tendency to purchase fewer apparel

products during holidays or weekends.
⚫ While a smaller group, businesses should still consider this pattern when planning
inventory and marketing strategies for these periods.
⚫ Demand forecasting models should be flexible enough to adjust for potential dips in
demand during holidays or weekends, allowing for more precise inventory management.

Consistent Purchases During Holidays or Weekends (No Difference):

⚫ The majority, with 46 respondents, reported no significant change in apparel purchasing

habits during holidays or weekends.
⚫ For this group, demand forecasting models should ensure stability in predicting regular
sales patterns during these periods, allowing for effective inventory planning.

Inference on Impact of Availability During Holidays or Weekends:

Perceived Impact on Availability (Yes):

⚫ A total of 42 respondents believe that the availability of apparel products is affected
during holidays or weekends.
⚫ This perception indicates a potential concern among consumers regarding the availability
of desired products during peak shopping times.
⚫ For businesses, this insight suggests the need to ensure robust inventory management
and supply chain strategies during holidays or weekends to meet the perceived demand
and avoid potential stock-outs.

Perceived No Impact on Availability (No):

⚫ A larger group of 61 respondents does not believe that the availability of apparel products
is affected during holidays or weekends.
⚫ This perception suggests that, according to this segment of respondents, businesses
effectively manage product availability during peak periods.
⚫ While this group may not alter their purchasing behaviour based on perceived availability
issues, businesses should still maintain rigorous inventory planning to meet demand
expectations and customer satisfaction.
Inference on Importance of Demand Forecasting for Apparel Products:

Low Importance (Scale 1-4):

⚫ A combined total of 21 respondents (11 at scale 1, 3 at scale 2, and 4 at scale 4) indicated

a relatively lower importance of demand forecasting for apparel products.
⚫ Individuals in this group may not perceive demand forecasting as a critical factor in their
purchasing decisions, suggesting that other factors such as price or style may take
precedence.
⚫ For businesses, this signals a need to emphasize the benefits of accurate demand
forecasting in improving inventory management and meeting customer expectations.

Moderate Importance (Scale 5-7):

⚫ A combined total of 21 respondents (5 at scale 5, 8 at scale 6, and 8 at scale 7) indicated

a moderate importance of demand forecasting.
⚫ This group recognizes the value of demand forecasting but may not consider it a decisive
factor in their purchasing decisions.
⚫ Businesses should ensure that their demand forecasting strategies are effective in
meeting the needs of this moderately sensitive group, emphasizing the benefits of
improved product availability and reduced stock-outs.
High Importance (Scale 8-10):

⚫ A combined total of 59 respondents (19 at scale 8, 13 at scale 9, and 27 at scale 10)

emphasized the high importance of demand forecasting for apparel products.
⚫ This group places a premium on accurate forecasting, suggesting that businesses need to
prioritize investments in advanced forecasting models and technologies to meet
customer expectations.
⚫ Effective demand forecasting is likely to influence the purchasing decisions of this highly
sensitive group, impacting their overall satisfaction with the shopping experience.
Inference on Perceived Changes in Demand During Holidays or Weekends:

Perceived Changes in Demand (Yes):

⚫ A total of 53 respondents believes that the demand for apparel products changes during
holidays or weekends.
⚫ This perception indicates an awareness among consumers that shopping behaviors may
be influenced by external factors such as holidays or weekends.
⚫ For businesses, this insight suggests the importance of incorporating seasonality into
demand forecasting models to accurately predict and meet shifts in demand during these
specific time-frames.

Perceived No Changes in Demand (No):

⚫ A group of 50 respondents does not believe that the demand for apparel products
changes during holidays or weekends.
⚫ This perception suggests that, according to this segment, consumer demand remains
relatively stable across different time-frames.
⚫ While this group may not perceive significant changes in demand, businesses should still
consider the potential impact of marketing initiatives or promotions during holidays or
weekends to capture any variations in consumer behavior.

Inference on Factors Influencing Apparel Purchasing Decision:

1. Price:
⚫ A significant number of respondents (21) consider price as the most influential factor in
their apparel purchasing decision.
⚫ Price continues to play a role in subsequent preferences, with varying importance across
different positions (2nd to 7th).
⚫ Businesses should prioritize competitive pricing strategies and consider promotions or
discounts to appeal to this segment.

2. Quality:
⚫ Quality is a key consideration for a substantial number of respondents, with 16
individuals placing it as the top influencing factor.
⚫ Similar to price, quality maintains importance across different positions, suggesting a
discerning consumer segment.
⚫ Businesses should emphasize product quality in marketing and product descriptions to
appeal to this group.

3. Brand:
⚫ Brand loyalty is evident, with 15 respondents placing it as the top factor and 21 as the
second most important.
⚫ This emphasizes the significance of brand reputation and recognition in influencing
apparel purchasing decisions.
⚫ Businesses should focus on building and maintaining a strong brand image to attract and
retain customers.

4. Style:
⚫ Style is a significant factor for a considerable number of respondents, with 13 placing it
as the top influencing factor.
⚫ Like other factors, style maintains relevance across different positions, indicating the
importance of staying on-trend.
⚫ Businesses should stay updated on fashion trends and offer diverse styles to cater to
various consumer preferences.

5. Comfort:
⚫ Comfort is an important consideration for respondents, with 17 individuals placing it as
the top influencing factor.
⚫ Similar to other factors, comfort maintains relevance across different positions,
indicating its sustained importance.
⚫ Businesses should prioritize comfort in product design and marketing to meet the
preferences of this segment.

6. Availability:
⚫ Availability is considered crucial by a significant number of respondents, with 19 placing
it as the top influencing factor.
⚫ This emphasizes the importance of maintaining a robust inventory and efficient supply
chain to meet consumer demand.
⚫ Businesses should focus on optimizing inventory management and communicate product
availability effectively.
Inference on Residency Country and Demand Forecasting:

Overall Context:
⚫ It's important to note that the survey was predominantly conducted in India, resulting in
a concentration of responses from this region.
⚫ The high number of Indian respondents underscores the significance of tailoring demand
forecasting models to the specific needs and preferences of the Indian market.
⚫ Businesses should consider the diverse cultural, climatic, and regional factors within India
to enhance the accuracy of demand forecasting.

OVERALL CONCLUSION:

The survey provides a comprehensive snapshot of the preferences and behaviors of

individuals primarily residing in India. With a significant majority of respondents (92) hailing
from India, businesses should prioritize tailoring their strategies to meet the specific needs of
this diverse and dynamic market. Understanding regional variations, cultural nuances, and
climatic influences within India is crucial for accurate demand forecasting, enabling businesses
to optimize inventory management and marketing strategies.
Insights from other countries, while limited in this survey, offer valuable glimpses into
consumer preferences in regions such as the United States. Though sample sizes for these
countries are smaller, businesses can glean insights for targeted strategies in these
international markets.
Key factors influencing apparel purchasing decisions, such as price, quality, brand, style,
comfort, and availability, showcase the multifaceted nature of consumer preferences.
Businesses should adopt a holistic approach, considering these factors in tandem, to develop
effective demand forecasting models and cater to diverse consumer segments.
The data on the frequency of apparel purchases, changes in purchasing behavior during
holidays or weekends, and perceived impact on product availability during these periods
highlight the need for businesses to adapt their strategies to meet fluctuating demand. This
flexibility is crucial for businesses to navigate seasonal variations and capitalize on peak
shopping periods.
REFERNCES:

1. Wiyanti, DT, et al. “Machine-learning algorithm for demand forecasting problem.” Journal of Physics:
Conference Series, vol. 1918, no. 4, 2021, p. 042012, https://doi.org/10.1088/1742-
6596/1918/4/042012.

2. Zohdi, Maryam, et al. “Demand forecasting based machine learning algorithms on Customer
Information: An applied approach.” International Journal of Information Technology, vol. 14, no. 4,
2022, pp. 1937–1947, https://doi.org/10.1007/s41870-022-00875-3.

3. Vasudev, Shekhar  Ramesh. Demand Forecasting Using Statistical and  Machine

Learning Algorithms, 2018.

4. Kochak Ashvin, and Suman Sharma. “DEMAND FORECASTING USING NEURAL NETWORK FOR SUPPLY
CHAIN MANAGEMENT.” International Journal of Mechanical Engineering and Robotics Research, vol. 4,
Jan. 2015.

5. Terrada, Loubna, et al. “Demand forecasting model using Deep Learning Methods for Supply Chain
Management 4.0.” International Journal of Advanced Computer Science and Applications, vol. 13, no.
5, 2022, https://doi.org/10.14569/ijacsa.2022.0130581.

6. Aamer, Ammar, et al. “Data Analytics in the Supply Chain Management: Review of Machine Learning
Applications in demand forecasting.” Operations and Supply Chain Management: An International
Journal, vol. 14, 2021, https://doi.org/10.31387/oscm0440281.

7. .K, Srinithi, et al. A Machine Learning System for Demand Forecasting Using Light Gradient Boosting
Machine Algorithm, vol. 6, no. 1, 25 Jan. 2023.

8. Kilimci, Zeynep Hilal, et al. “An improved demand forecasting model using Deep Learning Approach
and proposed decision integration strategy for Supply Chain.” Complexity, vol. 2019, 2019, pp. 1–15,
https://doi.org/10.1155/2019/9067367.
APPENDIX
11/13/23, 3:56 PM Shopping Habits and Product Availability

Shopping Habits and Product Availability

[email protected] Switch account
Draft saved
Not shared

* Indicates required question

E-mail ID

Your answer

What is your gender? *

Male

Female

Other

What is your age group? *

18-24

25-34

35-44

45-54

55+

https://docs.google.com/forms/d/e/1FAIpQLScolJeKfiurepmWjUsvu3Vr00_vEam7uh_iEhx1vt_2mqib3A/viewform 1/7
11/13/23, 3:56 PM Shopping Habits and Product Availability

Which of the following best describes your employment status? *

Student

Employed full-time

Employed part-time

Unemployed

Retired

Self-employed

How often do you purchase apparel products? *

Daily

Weekly

Monthly

Yearly

Rarely

Never

How important is product availability when purchasing apparel products? *

1 2 3 4 5 6 7 8 9 10

How do holidays or weekends affect your apparel product purchasing habits?

Your answer

https://docs.google.com/forms/d/e/1FAIpQLScolJeKfiurepmWjUsvu3Vr00_vEam7uh_iEhx1vt_2mqib3A/viewform 2/7
11/13/23, 3:56 PM Shopping Habits and Product Availability

During holidays or weekends, do you tend to purchase more or less apparel *

products than usual?

More

Less

No difference

Do you think that the availability of apparel products is affected during holidays or *
weekends?

Yes

No

If yes, how does it affect your purchasing decision? *

Your answer

How important is demand forecasting for apparel products? *

1 2 3 4 5 6 7 8 9 10

Do you think that demand for apparel products changes during holidays or *
weekends?

Yes

No

https://docs.google.com/forms/d/e/1FAIpQLScolJeKfiurepmWjUsvu3Vr00_vEam7uh_iEhx1vt_2mqib3A/viewform 3/7
11/13/23, 3:56 PM Shopping Habits and Product Availability

If yes, how does it affect your purchasing decision?

Your answer

Which of the following factors influence your apparel product purchasing *

decision?

1 2 3 4 5 6 7

Price

Quality

Brand

Style

Comfort

Availability

Others

https://docs.google.com/forms/d/e/1FAIpQLScolJeKfiurepmWjUsvu3Vr00_vEam7uh_iEhx1vt_2mqib3A/viewform 4/7
11/13/23, 3:56 PM Shopping Habits and Product Availability

In which country do you reside? *

USA

Canada

UK

Australia

Germany

France

Spain

Italy

Japan

China

India

Brazil

Mexico

Other

This is a required question

Submit Clear form

Never submit passwords through Google Forms.

This content is neither created nor endorsed by Google. Report Abuse - Terms of Service - Privacy Policy

Forms

https://docs.google.com/forms/d/e/1FAIpQLScolJeKfiurepmWjUsvu3Vr00_vEam7uh_iEhx1vt_2mqib3A/viewform 5/7