1
Table of Contents
I. Introduction ................................................................................................................................ 3
1. Abstract .................................................................................................................................................. 3
2. Introduction ......................................................................................................................................... 3
1. Flowchart .............................................................................................................................................. 5
III. The dataset.............................................................................................................................. 6
1. General overview of the data set .................................................................................................. 6
2. Reasons to use data sets .................................................................................................................. 7
3. Overview of machine learning models ....................................................................................... 7
1. Time Series Analysis: ........................................................................................................................ 7
2. Dashboard: ..........................................................................................................................................10
V. Data cleaning........................................................................................................................... 10
VI. Explore data analysis ( EDA ) ......................................................................................... 12
VII. Model ...................................................................................................................................... 19
1. K-Mean..................................................................................................................................................19
2. Linear Regression - Order Quantity Forecast ........................................................................23
VIII. Conclusion ............................................................................................................................ 26
1. Overview of project results ..........................................................................................................26
2. Proposal for strategic development ..........................................................................................26
3. Project future .....................................................................................................................................27
IX. Contribution ........................................................................................................................ 28
X. Reference................................................................................................................................... 28
2
I. Introduction
1. Abstract
The project "Customer Segmentation for Revenue and Inventory Management" focuses on
enhancing revenue and inventory management for DPR Construction, an international
construction company headquartered in Redwood City, California, USA. Established in 1990,
DPR Construction operates globally, specializing in construction and project management
across various sectors such as healthcare, industrial, education, and technology.
The project aims to address the challenges that DPR faces in today's business environment.
The integration of predictive data analysis and customer segmentation creates a comprehensive
strategy, enabling the company to optimize performance and adapt quickly to market
fluctuations in the construction industry. The project establishes the foundation for building an
intelligent sales and inventory management system, providing business flexibility and
minimizing risks. Through this approach, DPR Construction can swiftly address issues,
optimize resources, and achieve stable success in the global construction market.
2. Introduction
In today's modern business world, revenue and inventory management are crucial to the
success of any organization and business. Mastering the sales process, revenue, profit and
inventory management is not only a decisive factor in strong competitiveness but also plays a
role in determining success or failure in today's market. Our project focuses on classifying
potential customers for a Construction Company, using advanced methods including customer
segmentation through K-means clustering, Order Quantity Forecasting using Linear Regression
and Time Series Analysis to understand customer shopping trends, the project strives to provide
actionable information.
We classify customers based on construction needs, project size and budget to optimize
marketing and sales strategies. Our goals extend beyond improving sales and inventory
management; We look forward to building long-term partnerships. Through this project, we
hope to establish strong relationships, while enhancing the competitiveness and stability of the
Construction Company in a challenging market.
II. Project idea
1. Project idea
3
In today's business environment, effective inventory and sales management is crucial to
the success of any organization or business. A deep understanding of the sales process, revenue,
profit, and inventory management can be the deciding factor between competitive success and
failure in the marketplace.
Our project aims to apply methods like Linear Regression and Time Series Analysis to
forecast order quantity for a Construction Company. We also incorporate K-means customer
segmentation to create a comprehensive sales and inventory management strategy.
Detail goal:
Order Quantity Forecast:
Use Linear Regression to forecast expected order volume based on factors such as project
size, retail price, and order history.
Customer Classification:
Use K-means to classify customers into segments based on construction needs, project size,
and budget.
Effective Inventory Management:
Optimize inventory management based on forecasts and characteristics of each customer
segment. Use Time Series Analysis to gain insights into customer buying trends and cycles.
Optimize Sales Strategy:
Apply data from forecasting and classification to optimize sales strategies, helping
businesses adapt flexibly to the market.
Project planning and scheduling:
Applying forecasting and classification also help project planning more flexibility to deal
with problems that can happen, companies can enhance their project management, improve
resource utilization, reduce risks, and ultimately optimize the overall efficiency and success of
construction projects.
Hope:
We hope that through this project, we will not only provide accurate forecasts of order
volume, but also help the company see how to build a smart sales and inventory management
system. The project represents the integration of predictive data analytics and customer
segmentation to create a comprehensive strategy, helping businesses optimize performance and
adapt quickly to construction market fluctuations.
4
1. Flowchart
Start:
The project started with setting the main goal: Classifying
potential customers for the Construction Company.
Explore Data:
Research and understand data from DPR Construction
Company's data set.
Review data structure, check for missing values, and create an
overview of customer and transaction information.
Data Cleaning:
Perform data cleaning steps to normalize and remove noise.
Handle missing values and make data ready for analysis.
EDA (Exploratory Data Analysis):
Conduct detailed data analysis to better understand the
relationships between variables and trends in customer behavior.
Use charts and graphs to describe information and uncover
important insights.
Forecast Order Quantity:
Use predictive modeling to estimate future orders based on
historical trends and scenario projections.
Customer Segmentation:
Classify customers into groups based on construction needs,
project size, and budget.
Identify key characteristics that help optimize marketing and
sales strategies.
Inventory Management:
Inventory management is based on information from EDA and
forecasts, to ensure efficient response to demand and reduce the risk
of loss.
Optimize Sales Strategy:
Optimize sales strategy based on customer analysis and order
forecasting. Create marketing and sales strategies suitable for each
customer group.
Tableau & Dashboard:
Use Tableau to create dashboards and data visualizations.
Display information clearly and easily for strategic decision-
making.
End:
At the end of the project and drawing conclusions and solutions
through analysis, the Construction Company will achieve a good and stable competitive
landscape in the construction market.
5
III. The dataset
1. General overview of the data set
This data was collected from DPR Construction consisting of sales order information,
detailing transactions between a company (presumably a supplier or manufacturer) and various
customers. Each sales order entry includes key information such as the order number, order
date, customer name, customer type (e.g., Distributor, Wholesale, Export), warehouse code,
product code, order quantity, unit price, revenue generated, and associated costs. There are 8000
rows and 10 columns, each column has a specific function:
Column Name Description Data Type
Unique identification code for each
OrderNumber object
transaction.
OrderDate The date on which the order was placed. object
CustomerName The name of the customer placing the order. object
Classification of the customer (e.g., retail,
Customer Type object
wholesale, export).
Code of the warehouse from where the
WarehouseCode object
product is shipped.
Product Code Identification code of the ordered product. object
The quantity of the product ordered in each
OrderQuantity int64
transaction.
UnitPrice Retail price per unit of the product. float64
Revenue Total revenue generated from the order. float64
int64
Costs Costs associated with the order.
6
2. Reasons to use data sets
By analyzing this dataset, we can gather valuable insights to develop a marketing strategy
for a construction company. Factors such as customer segmentation, product performance,
order trends, customer behavior, costs, and warehouse efficiency provide key information about
customers, products, and business processes. Utilizing these insights can inform decisions on
pricing, advertising, and focusing on specific customer groups, ultimately optimizing
profitability and overall business performance.
3. Overview of machine learning models
K-means is an unsupervised clustering algorithm designed to partition unlabelled data into
a certain number (that's the “ K”) of distinct groupings. In other words, k-means finds
observations that share important characteristics and classifies them together into clusters. A
good clustering solution is one that finds clusters such that the observations within each cluster
are more similar than the clusters themselves.
Linear regression is a statistical method used for modeling the relationship between a
dependent variable and one or more independent variables by fitting a linear equation to
observed data. The main goal of linear regression is to find the best-fit line that represents the
linear relationship between the variables.
Time series analysis focuses on studying and interpreting a sequence of data points
recorded or collected at consistent time intervals. Unlike cross-sectional data, which captures a
snapshot in time, time series data is fundamentally dynamic, evolving over chronological
sequences both short and extremely long. This type of analysis is pivotal in uncovering
underlying structures within the data, such as trends, cycles, and seasonal variations.
IV. Tableau & dashboard:
1. Time Series Analysis:
Time series analysis is a method of examining a set of data points collected over a period
of time. Time series analysis involves analysts recording data points at constant intervals over
a predetermined length of time rather than just occasionally or arbitrarily.
Reason why we do Time Series Analysis in Tableau is because we want to visualize data
in the most understandable and visible way possible. Using colored charts and graphs allows
you to clearly examine and see the data's properties, making it easier to exploit.
- Forecasting Order Quantity with time series analysis:
7
Using Tableau to do time series analysis for Order Quantity, we clearly see that our graph
visualizes the most orders appearing in August with 4,239 orders and followed after January
with 4,001 orders.
● To forecast we determine that our graph follows the ‘trend additive’, after applying we
have the MASE is 0.74 which means our estimate is good quality.
● Estimate will start from July 2023 to March 2014.
- Forecasting Revenue with time series analysis:
Using Tableau doing the same with forecasting Revenue. Due to the highest order in
August, August has the highest revenue with 1,398,499. After that is January with 1,335,340.
● Also use trend additive to estimate with 95% confidence, and we got the MASE is 0.76
which means our quality is good enough.
● Estimate from July 2023 to March 2024 in increasing value.
- Forecasting unit price with time series analysis
8
Using Tableau doing the same with forecasting Unit Price of product. The unit price was
also affected by the high order in August so the highest unit price appeared in August with
163,882.
● also using trend additive to forecasting the unit price from July 203 to March 2024. we
got the MASE is 0.69 which means good quality.
- Forecasting cost with time series analysis:
Apply Tableau to forecasting the Cost to see if the highest order is affected by the cost or
not. The answer is Yes. August still has the highest cost with 780,476.
● We still apply the Trend additive to estimate the cost from July 2023 to March 2024.
After applying we have a MASE of 0.78 which means good quality.
Time series analysis can be used to find patterns, seasonality, and trends in historical data
from a project. Like in our dataset, we can identify our dataset is due to trend, do we can forecast
better. This data can be used to create more precise project timetables, allowing for improved
resource allocation and on-time completion. Forecasting costs based on past spending habits.
This enables for more accurate budgeting and better financial planning throughout the project
lifespan. Construction organizations can foresee delays by reviewing historical performance
data, allowing for proactive risk mitigation techniques.
9
2. Dashboard:
V. Data cleaning
To understand more about the dataset we will use ‘[Link]()’.
The dataset consists of 10 columns and 8000 rows. So, 8000 non-null entries for all
columns indicate the absence of missing values (null). The data types of each column are also
displayed, including object, int64, and float64.
10
The descriptive statistics table provides an overview of key features in the dataset
We utilized [Link](inplace=True) to eliminate rows containing missing values from
our DataFrame, data. This ensures data cleanliness and assures that we are working with
complete information throughout the analysis.
● Outlier Checking:
We can clearly see outliers appear in Revenue.
- Z-score method:
We use the Z-score method to calculate upper limit and lower limit to easy to identify and delete
outlier.
After remove outlier, we have new data with 7957, remove 43 outlier from the dataset.
11
VI. Explore data analysis ( EDA )
1. The distribution of variables in the data set
● Order quantity fluctuates with peaks at 6 and 10. This may indicate that customers tend
to buy products with these quantities.
● Unit price has a left-skewed distribution, meaning that most orders have a low unit price.
This may affect the profit margin of the company.
● Revenue and costs also have left-skewed distributions, meaning that most orders have
low revenue and costs. This may suggest that the company is selling low-value products or has
low production costs.
2. Distribution of Revenue by Customer Name
This graph shows the distribution of revenue by customer name. Each bar represents a
different customer, and the height of the colored segments within each bar indicates the amount
12
of revenue from that particular source or category. The black dots above the bars represent
outliers or specific data points. Here are some highlights of the chart :
● The customer with the highest revenue is MARTIN21, with over 7000 units of revenue.
The revenue of this customer mainly comes from two sources: green and blue.
● The customer with the lowest revenue is WHITE Co Ltd, with only about 500 units of
revenue. The revenue of this customer only comes from a single source: red.
● The black dots above the bars indicate outliers, meaning that the revenue values are very
high or very low compared to the normal distribution. For example, the customer WILSON Co
Ltd has a black dot at the revenue level of about 6000, while most of the revenue of this
customer is below 2000.
3. Oder count by Customer Types, WareHouse , Product
● The customer segment with the highest number of orders is Wholesale, accounting for
approximately 4000 orders. This may indicate that the company is more focused on the
wholesale market than other markets.
● On the other hand, the customer segment with the lowest number of orders is Export,
with only around 1000 orders. This could suggest that the company is facing challenges in
expanding into international markets or encountering increased competition there.
● The Distributor customer segment has an average order quantity, with over 2000 orders.
This may indicate that the company is maintaining a stable relationship with its distribution
partners.
13
This chart shows the order count by warehouse. Warehouse AXW291 has the highest order
count, exceeding 3500 orders. Warehouses NXH382 and GUT930 have significantly fewer
orders, with counts around 1000 and 1500 respectively. Warehouse FLR025 has the lowest
order count, below 500
This chart shows the order count by product. Each product is represented by a product code
and has multiple colored bars indicating different types or categories within that product. The
order count is shown on the y-axis, from 0 to 140.
● The highest bar in the chart is the orange bar STK - 146 and STK - 136 with an order
count of about 130. This shows that this product is very popular and has high demand in the
market.
14
● The lowest bar in the chart is the red bar STK - 157 , with an order count of only about
[Link] shows that this product is not very interested and has low demand in the market.
● The highlight in the chart is the large disparity in order counts among different products.
It can be seen that some products have very high order counts, while some others have very low
ones. This may reflect the difference in quality, price, marketing, or customer satisfaction with
these products.
4. Relationship between Revenue with Oder Quantity, Costs
● The number of orders varies between 5 and 12, while the revenue ranges from 0 to 8000.
● There is a clear upward trend; as the number of orders increases, revenue also increases,
despite significant fluctuations at each order quantity level.
15
● The costs range from 0 to 3500, while the revenue ranges from 0 to 8000.
● The majority of data points are concentrated at the lower end of both axes, indicating
that in many cases, both costs and revenue are low.
● As we move to the right along the x-axis (increasing costs), there is often an observed
upward trend on the y-axis (increasing revenue).
➔ In conclusion , both order quantity and costs appear to have a positive impact on
revenue.
5. Compare between Revenue and Costs in OderDate
● Both revenue and costs fluctuate over time, with peaks and troughs visible.
● Revenue consistently remains above costs throughout the entire timeline.
● There are some order dates where the revenue and costs have a large gap, indicating a
high profit margin.
● There are some order dates where the revenue and costs are close together, indicating a
low profit margin.
➔ The company consistently achieves higher revenue than costs, showcasing strong
business efficiency. There is a positive correlation between costs and revenue, indicating that
increased investment in activities like production, marketing, or research and development
tends to boost revenue. However, significant fluctuations in both costs and revenue on different
order days suggest business operations may be influenced by external factors like competition
or market dynamics.
6. Compare between Total Revenue of Warehouse with Product Code, Customer Type
16
● The customer segment generating the highest revenue is Distributors, with
approximately 2700 currency units, predominantly contributed by warehouse NHX382 (around
1200 currency units).
● The customer segment with the lowest revenue is Export, totaling about 900 currency
units, with warehouse FLR025 contributing the least (approximately 100 currency units).
● The warehouse with the highest revenue is NHX382, generating around 2000 currency
units, predominantly driven by the Distributor customer segment (about 60%).
● The warehouse with the lowest revenue is FLR025, yielding about 600 currency units,
with the Export customer segment contributing a smaller portion (around 17%).
17
● NXH382 has the highest revenue, followed by GUT930 and AXW291, FLR025
● High revenue tends to concentrate on specific warehouse codes, such as NXH382.
● Some product codes are top sellers across all warehouses, for example, STK-139, STK-
112.
● Certain product codes perform well only in specific warehouses, like STK-101, which
is popular only in NXH382.
➔ In summary, the analysis of revenue across warehouse codes, product codes, and
customer types has highlighted significant relationships crucial for business strategy.
Warehouse NXH382 plays a pivotal role in generating high revenue, primarily from Wholesale
customers. Products like STK-139, STK-112 contribute substantially to the overall revenue,
while some products perform exceptionally well only in specific warehouses, emphasizing the
diversity in product performance. The Wholesale customer type dominates the revenue share,
while the Export customer type exhibits the lowest revenue.
7. Correlation Heatmap
● The quantity of orders exhibits a perfect positive correlation with itself (1.00), but no
correlation with the unit price (0.00).
18
● Revenue shows a positive correlation ranging from moderate to strong with all variables
except for the quantity of orders.
● Costs also exhibit a positive correlation ranging from moderate to strong with all
variables except for the quantity of orders.
VII. Model
1. K-Mean
Elbow Method: Is a technique used to determine the optimal number of clusters in the K-
Means model. It reflects the change of WCSS (Within-Cluster-Sum-of-Squares) as the number
of clusters increases. A sudden decrease in WCSS creates an elbow-like shape on the graph,
and where this elbow appears is often chosen as the optimal number of clusters.
K-Means Clustering: Is a clustering algorithm in machine learning. It groups data points
into clusters such that the sum of squared distances from each point to the nearest cluster center
is minimum.
In this project, Elbow Method and K-Means are used to classify the Company's potential
customers. Elbow Method is applied to select the optimal number of clusters in K-Means,
helping to optimize clustering of customer data based on construction needs, project scale, and
budget.
Use a loop to test the number of clusters from 1 to 10.
We can see, the "elbow" point is at the number of clusters 3, meaning here, 3 clusters is
the best choice in the data.
19
Create a new K-Means model with a cluster count of 3 and classify the data into clusters
and assign a cluster label to each data point.
Classification chart based on two features ('OrderQuantity' and 'Revenue')
The classification chart is based on two features ('OrderQuantity' and 'Revenue')
20
Classification chart based on 'UnitPrice' and 'Costs' features
21
Classification chart based on all three features ('OrderQuantity', 'Revenue', 'Costs')
We see that each order has been classified into one of 3 clusters (0, 1, 2)
Cluster 0 (Cluster 0): Orders with OrderNumber of "SO - 000103". Large OrderQuantity (10),
high UnitPrice (643.50), leading to high Revenue and Costs.
This is an order from a customer with great demand, ordering in large quantities and at a high
price.
Cluster 1 (Cluster 1): Orders have OrderNumbers "SO - 000104" and "SO - 000105".
OrderQuantity and UnitPrice are average, resulting in Revenue and Costs being average.
This is an order from a customer with average needs and budget.
Cluster 2 (Cluster 2): Orders with OrderNumber are "SO - 000101" and "SO - 000102". High
OrderQuantity and UnitPrice lead to high Revenue and Costs.
This is an order from a large customer, ordering in large quantities and at a high price.
Thereby, we can see that the majority of customer groups with great demand are often
concentrated in clusters 0 and 2, that is, investors who need large quantities and have high
revenue. There are significant differences between clusters in terms of purchase levels and order
values between clusters 1 and 0; 2 provides a clear understanding of the characteristics of each
22
group that can help optimize marketing, inventory management, and customer engagement
strategies while also providing a fine-tuned business strategy to reflect the characteristics and
different requirements of each customer group.
2. Linear Regression - Order Quantity Forecast
This section aims to make effective forecasts by developing an inventory forecasting
model using Linear Regression algorithm to ensure forecast accuracy and efficiency. Support
inventory management by providing detailed information on resource demand forecasts. This
helps optimize inventory and avoid shortages and waste. Ensure that the company uses
resources optimally through accurate forecasting of revenue from orders and associated costs.
The main goal of this analysis is to develop a model that can predict order quantities based on
key financial metrics. This will help the company better predict material requirements, optimize
inventory levels, and ensure timely project execution.
Additionally, accurate forecasting enhances business performance by eliminating
uncertainty and reducing risk when managing inventory. Forecast-based inventory management
helps minimize waste due to overstocking or loss due to shortages. Accurate forecasting of
customer needs helps the company respond promptly and flexibly, increasing customer
satisfaction.
The analysis process begins with preparing the data set, which includes important financial
indicators such as Unit Price, Revenue, and Expenses as independent variables. These variables
are found to have a significant influence on order quantity, thus forming the basis of our
prediction model.
23
Next, we divided Data and Model Training. Create and train a linear regression model
with the training data set. This model aims to find a linear relationship between independent
and dependent variables. To establish the relationship between independent variables (Unit
Price, Revenue, and Cost) and dependent variables (Order Quantity). The dataset is divided into
training and testing subsets to validate the model's performance. This is important for evaluating
the model on unseen data, which helps test the generality of the model.
Model Accuracy
In the next step, we evaluated the Model using metrics like R-squared, MSE, and RMSE.
The R-squared index reflects the percentage of variation in 'OrderQuantity' explained by the
model. Mean Squared Error, measures the average of the squares of the errors. Root Mean
Squared Error, which is the square root of MSE, provides an estimate of the standard error. The
found R-squared (R²) values are 0.866, showing that the model is capable of explaining about
86.6% of the variation of the variable 'OrderQuantity'. This is a good result, showing that the
model fits the data quite closely. MSE and RMSE have values of 0.699 and 0.836, respectively.
RMSE provides an estimate of the standard error of predictions relative to the actual value, and
in this case, it is relatively low.
These indicators show that the linear regression model performs well in predicting the
order quantity based on the selected independent variables.
24
The scatter chart compares the predicted value (Predicted OrderQuantity) and the actual
value (Actual OrderQuantity) of the order quantity. The dashed red line represents "Perfect
Prediction", where the prediction exactly coincides with the actual value. The histogram allows
us to visually see the distribution of predicted values compared to actual values. The closeness
of the points to the "Perfect Prediction" line indicates that the prediction model is relatively
accurate.
The scattered points show which specific cases the model predicts well and which cases
have large errors. This helps identify bias and can provide guidance on how to improve the
model. Any departure from the perfect prediction line may indicate an outlier or data point that
does not follow the general trend. This can be useful for detecting anomalies in order processes
or data. If the model shows high accuracy, it can be used to predict future order quantities based
on factors such as price and cost, helping with effective planning and inventory management.
more effective.
Predicted Results
The model is then used to predict order quantities based on new sample data. This step
is important to understand the practical application of the model and its effectiveness in real-
life situations. We have created a set of predicted order quantities based on certain financial
metrics. These predictions are instrumental in helping companies make informed decisions
about inventory and supply chain management.
Each row in the table shows a prediction of the order quantity based on the corresponding
unit price, revenue, and costs. The "Predicted OrderQuantity" value ranges from a lowest value
of about 2.67 to a highest value of about 17.22.
Forecasts can be categorized by customer groups to identify different demand and
ordering behaviors across segments. Forecasting can assist in supply planning and inventory
management by providing information about how much product needs to be produced or
purchased. Compare forecasted volumes with actual results to evaluate the effectiveness of
current sales strategies.
Depending on the actual situation, businesses can adjust price parameters and desired
revenue to find the optimal and most appropriate order quantity to consider. To ensure market
demand and avoid backlogs due to too much stockpiling, or lack of goods that are not enough
to meet customer needs.
25
VIII. Conclusion
1. Overview of project results
In summary, through this report, we have successfully solved the problems to
significantly improve sales efficiency and inventory management. Our core methods for solving
these problems include customer segmentation using K-means clustering, order quantity
forecasting through Linear Regression, and effective inventory management systems.
Our project's foremost accomplishment is the development of a sophisticated customer
segmentation model using K-means clustering. This model successfully identified distinct
customer groups based on purchasing behavior, geographic location, and order history. This
segmentation enables the company to tailor its marketing and sales strategies more effectively,
ensuring a more personalized approach to client interactions.
Through the implementation of Linear Regression analysis, we achieved a significant
improvement in forecasting the quantity of orders. This predictive capability allows for more
accurate stock management and order fulfillment, reducing both overstocking and stockouts.
The forecasting model takes into account historical sales data, market trends, and seasonal
variations, providing a robust and reliable prediction tool.
The integration of customer segmentation and order forecasting into the inventory
management system has revolutionized how inventory is managed. The system can now
dynamically adjust inventory levels based on customer demand forecasts, leading to optimized
stock levels, reduced storage costs, and minimized wastage.
The project has empowered the company to make more data-driven decisions. The
insights derived from the clustering and forecasting models provide a deeper understanding of
the market, enabling the company to adapt quickly to changing customer needs and market
conditions.
By understanding customer needs better and managing inventory more efficiently, the
company has seen an improvement in customer service levels. Faster response times,
personalized service offerings, and efficient handling of orders have all contributed to increased
customer satisfaction.
2. Proposal for strategic development
To take advantage of the project's research, the strategic development plan should focus
on further improving predictive analytics capabilities. The company should consider expanding
the data set and incorporating additional predictive variables to improve the accuracy of
customer segmentation and order forecasting.
The company should explore the use of more sophisticated machine learning algorithms
and AI-driven analytics to enhance predictive accuracy and gain deeper insights into customer
behavior and market dynamics.
Allocate resources to adopt cutting-edge AI and machine learning technologies, which
can automate complex data analyses and provide real-time insights. Utilize the insights from
customer segmentation to design targeted marketing campaigns, focusing on the specific needs
and preferences of different customer groups. Implement dynamic pricing strategies based on
customer segment, purchase history, and market demand to maximize profitability and
26
competitive edge. Further refine the inventory management system by integrating real-time
sales data and predictive analytics, ensuring optimal stock levels and reducing operational costs.
In addition, The company needs to improve the customer segmentation and forecasting
models by incorporating more diverse data sources, such as social media analytics, economic
indicators, and construction industry trends. Invest in training programs to improve the data
literacy of staff across departments, ensuring that all employees can effectively utilize data-
driven tools and insights in their work. Long-term strategic planning explores opportunities for
applying the project's methodologies and tools to new markets or regions, considering both
domestic and international expansion. Establish a culture of continuous improvement, regularly
updating models and strategies based on market feedback and technological advancements.
3. Project future
The project has created a solid foundation for the Construction Company's development
in a competitive market. The next phase will involve scaling the application across different
departments and integrating it with other business processes for a more unified approach.
Continued investment in technology and employee training will be important to adapt to
evolving market trends. Furthermore, fostering a culture of innovation and data-driven
decision-making will be key to maintaining a competitive advantage. The ultimate goal is to
make the company not just a construction business but also a forward-thinking business that
uses industry-leading technology.
Finally, we would like to take a moment to express our sincere gratitude to lecturer Dr.
Ha Manh Hung. Your insights and insights into complex analytical processes not only provide
us with the tools needed to achieve our goals but also enrich our learning experience. The
knowledge and skills we gain under your guidance will certainly be valuable assets for our
future endeavors.
Thank you Dr. Ha Manh Hung for your dedication and key role in our academic and
professional development. We are grateful for the opportunity to work under your guidance and
look forward to applying the lessons learned to our future projects and careers. Your mentorship
was not only insightful but also inspiring, shaping our approach to complex challenges. We
appreciate your dedication and support throughout this journey.
27
IX. Contribution
Student name Student Contribution
ID
Vũ Diệu Huyền 20070938 + Model: Linear Regression
+ Overview of project results
+ Proposal for strategic development
+ Project future
+ Supports code EDA and General
overview of the data set
Tạ Thị Kim Hân 20070926 + Model: Time Series Analysis
+ Tableau
+ Project idea
+ Support data cleaning
Trần Hương Quỳnh 20070975 + Team leader
+ Model: K-Mean
+ Project idea
+ Flow chart
+ Build outline andCheck out the
entire content
Trương Trà My 20070960 + Edit code and explain EDA
+ General overview of the data set
+ Reasons to use supply chain & data
sets
+ Overview of machine learning
models
Trần Thu Hoài 20070932 + Data cleaning
+ Abstract
+ Introduction
+ Edit report
X. Reference
[1] DeepAI. (2019). K-Means. [online] Available at: [Link]
glossary-and-terms/k-means.
[2] Statistics Solutions (2013). What is Linear Regression? [online] Statistics Solutions.
Available at: [Link]
analyses/what-is-linear-regression/.
Time Series Analysis: Definition, Types & Examples | Sigma Computing. [online]
Available at: [Link]
28