I NDIAN I NSTITUTE OF T ECHNOLOGY , M ADRAS

B USINESS DATA M ANAGEMENT P ROJECT

M ID T ERM R EPORT

Optimizing Procurement of Goods for a

Food-based FMCG Company

Submitted by:
ANHAT SINGH
21 F 2000381

May 9, 2023
Contents
1 Executive Summary 2

2 Proof of Originality of Data 2

3 Metadata and Descriptive Statistics 3

3.1 Descriptive Analysis of Date . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.2 Descriptive Analysis of Amount . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.3 Descriptive Analysis of SKUs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

4 Detailed Explanation of Analysis Process 6

4.1 Best way to optimize procurement of goods . . . . . . . . . . . . . . . . . . . . . . 6
4.2 SKU Pareto Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
4.3 Festivals vs Sales Correlation Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 7
4.4 Seasons vs Sales Correlation Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 8

5 Results and Findings 8

5.1 SKU Pareto Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
5.2 Festivals vs Sales Correlation Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 8
5.3 Seasons vs Sales Correlation Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 9

6 Conclusion and What next? 9

List of Figures
1 Descriptive Stats of Amount . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2 Share of individual company in the Bill of Sales . . . . . . . . . . . . . . . . . . . . 4
3 Number of units sold of each SKU in FY 2020-2022 . . . . . . . . . . . . . . . . . 6
4 Python Code to find correlation between festivals and sales . . . . . . . . . . . . . . 7
5 Correlation Heatmap between Festivals and Sales . . . . . . . . . . . . . . . . . . . 8
6 Trend line between Festivals and Sales . . . . . . . . . . . . . . . . . . . . . . . . . 9
7 Monthly Number of SKUs sold for the Top 5 SKUs . . . . . . . . . . . . . . . . . . 9

List of Tables
1 Documents to show the originality of the data . . . . . . . . . . . . . . . . . . . . . 2
2 Description of columns in the dataset . . . . . . . . . . . . . . . . . . . . . . . . . . 3
3 Various SKUs offered by the company . . . . . . . . . . . . . . . . . . . . . . . . . 5
 Optimizing Procurement of Goods for a Food-based FMCG Company

1 Executive Summary
The document at hand presents a comprehensive analysis of a dataset, encompassing proof of origi-
nality, descriptive statistics, analytical methodology and findings. The dataset comprises
information on sales and purchases for two consecutive financial years, from April 2020 to March
2022. The data was collected from Bills of Sales and Purchases, and a total of 1014 Bills of Sales
and 392 Bills of Purchases were obtained.

The descriptive statistics reveal insights into the dataset, including the top ten companies that
procure goods from the and the distribution of companies that purchase the products. The
analytical process involved correlation analysis using Pandas and monthly analysis performed using
MS Excel to investigate whether there is a positive correlation between festivals, seasons and sales.

The results and findings of the analysis indicate a significant positive correlation between seasons
and sales and no correlation between festivals and sales. Consequently, it is essential to ensure that
there is an adequate inventory of units available to meet the anticipated demand during the various
seasons. Proper inventory management and forecasting can help optimize sales and profits while
preventing stockouts or overstocks during peak demand periods.

Overall, the document provides valuable insights into the dataset, enabling businesses to make in-
formed decisions regarding inventory management and sales optimization during various seasons.

2 Proof of Originality of Data

As soon as the preliminary discussions were completed with a representative of the company, data
was collected. I discussed the project with the co-owner
who highlighted the problems being faced in fulfillment of orders due to untimely
procurement of goods. The sales and purchase data was available in the form of Bills of Purchases
and Bills of Sales. In addition to the Bills, a list of number of Stock Keeping Units (SKUs) sold
yearly was also collected. The Bills and SKUs information was converted to digital format as Excel
Sheets. The company keeps no inventory. The procurement and processing of raw materials is done
upon receiving new orders.

Name Type Link

Official Documents
A Letter bearing the company’s letterhead permitting the use of
PDF
the data
Sample of a Bill having the company name Images
A Video capturing a brief discussion of the problem statement
Video
with the co-owner of the company
After Data Collection
A compilation of all the Bills of Sales Google Sheet
A compilation of all the Bills of Purchases Google Sheet
A list of all the SKUs offered by the company Google Sheet
Sales data of the Top 5 SKUs Google Sheet
Table 1: Documents to show the originality of the data

Page 2 of 9
 Optimizing Procurement of Goods for a Food-based FMCG Company

3 Metadata and Descriptive Statistics

After detailed discussion with the co-owner, Excel Sheets containing the Bills of Sales, Bills of Pur-
chases, List of SKUs and Bills of Top 5 SKUs have been compiled. Table 2 highlights the details of
each feature in the dataset.

Column Name Datatype Description

Bills of Sales and Purchase
Date Date Date when the bill was issued
Bill No Categorical Unique number to identify each bill
Type Categorical Indicates whether the Bill is of Purchase or of Sale
Bill of Purchase contains the company from which raw material
Account Categorical is bought. Bill of Sales contains the company to which the end
product is sold.
Amount Numerical The total amount on the Bill
SKUs Data
SKU ID Categorical Unique ID to identify each SKU
Name Categorical Name of the SKU
The year the SKU was first introduced by the company to the
Year Introduced Date
target market
Unit Weight Numerical Weight of a single unit of the SKU in grams
Selling Price Numerical Maximum Retail Price of a single unit (Rs. per unit)
The number of months for which the SKU can be safely con-
Shelf Life Numerical
sumed
Units sold Numerical The number of units of the SKU sold in the years 2020 to 2022
Additional information regarding the SKU, ingredients of the
More Info Categorical
SKU or the benefits of the SKU.
Top 5 SKUs Data
Date Date Date when the bill was issued
Bill No Categorical Unique number to identify each bill
SKU2 Numerical The quantity sold (in Kgs) of the SKU item 2
SKU6 Numerical The quantity sold (in Kgs) of the SKU item 6
SKU10 Numerical The quantity sold (in Kgs) of the SKU item 10
SKU17 Numerical The quantity sold (in Kgs) of the SKU item 17
SKU12 Numerical The quantity sold (in Kgs) of the SKU item 12
Table 2: Description of columns in the dataset

As per the problem statement, Date, Amount, and Top SKUs data is important to see a seasonal
trend.

Page 3 of 9
 Optimizing Procurement of Goods for a Food-based FMCG Company

3.1 Descriptive Analysis of Date

In a bill, dates typically refer to the specific time periods during which certain provisions or actions
take effect. These dates are often crucial to the implementation and enforcement of the bill’s provi-
sions and may be included in various sections throughout the bill. For example, a bill might specify
the date on which a tax increase goes into effect, or the date by which a certain agency must
complete a report or take a particular action. As the data obtained originates from Bills of Sales and
Purchases, the dates captured represent the issuance date of the respective bill. Based on the
collected data, the subsequent observations have been recorded:
Key Points
In the entire dataset, the date is of the format dd-mm-yyyy.
The dataset spans two consecutive Financial Years, specifically, from April 2020 to March 2022.
Within this timeframe, a total of 1014 Bills of Sales and 392 Bills of Purchases were acquired.
The project proposal included information on the precise dates on which the respective bills were
generated. It is important to note that the dates contained within the dataset may exhibit repetition,
given that the issuance dates of multiple bills can coincide.

3.2 Descriptive Analysis of Amount

The amount in a bill refers to the monetary value assigned to the goods or services being sold or
provided. It represents the total cost that the buyer is expected to pay to the seller or service
provider.

Figure 1: Descriptive Stats of Amount

Based on the order value, the goods procured need to be adjusted, so the distribution of top organi-
zations due to which Sales happen is important. The Figure 2 provided lists the Top 10
organizations that purchases goods from

Figure 2: Share of individual company in the Bill of Sales

Page 4 of 9
 Optimizing Procurement of Goods for a Food-based FMCG Company

Key Points
48% of the sales were generated from orders by

An average order is of Rs. 18,584. So, at least twice of this much capital should always be kept to maintain a

3.3 Descriptive Analysis of SKUs

It is a unique identifier assigned to a particular product or item in a retailer’s inventory management
system. SKUs are used to track inventory levels, sales, and restocking needs. They can be alphanu-
meric codes, bar codes, or any other identifying system that allows the product to be easily
recognized and tracked. SKUs help retailers efficiently manage their inventory, enabling them to
optimize inven- tory levels and prevent stockouts or overstocks.

Table 3: Various SKUs offered by the company

Key Points
The company offers 24 different SKUs (refer ). Many SKUs were introduced in various years from 2018 to 2022
The older the SKU, the more popular it is.
Upon discussions with the company representative, since almost all of the SKUs have different raw materials, op

Page 5 of 9
 Optimizing Procurement of Goods for a Food-based FMCG Company

4 Detailed Explanation of Analysis Process

4.1 Best way to optimize procurement of goods
Upon successive discussions with the company representative, it was concluded that the best way to
optimize the Procurement of goods was to create a proper inventory system for the company. It was
further discussed that since the raw material required to manufacture the end product does not have
a good shelf life, it is better to keep stocks of end-products (SKUs). Keeping a proper inventory will
lead to timely procurement of raw materials and thus optimizing the procurement of the goods.

As per the original hypothesis, the procurement of goods and thus the sales may be affected by
festivals and various seasons in the year. If the hypothesis is true, a preliminary inventory can be
made based on Festival dates and the seasons.

To analyze if a correlation exists, the following shall be done:

1. SKU Pareto Analysis- Identifying the key SKUs that account for 85-95% of the total sales.

2. Festivals vs Sales Correlation Analysis- Using Python-Pandas and MS Excel to check if

Fes- tival and Amount have a correlation.

3. Seasons vs Sales Correlation Analysis- Using MS Excel to check if Seasons and Quantity of
SKUs sold have a correlation.

4.2 SKU Pareto Analysis

Many SKUs have recently been introduced by the company and are not much popular in the market,
in comparison to the traditional SKUs sold by the company. After detailed talks with the company
representative, I came to the conclusion that it is very challenging to keep inventory of all the
SKUs, especially considering that the company does not have any inventory management as of yet.

So, the goal was to identify Top SKUs that could account for 85-95% of the sales. The inventory
shall be made to include only these SKUs. Data was already available in the form of Units Sold in
FY 2020-2022 in the SKUs dataset. A pivot table including the SKUs and the Units sold was
created. The Pareto trend line was analyzed to find the SKUs that cover the required number of
sales.

Figure 3: Number of units sold of each SKU in FY 2020-2022

Page 6 of 9
 Optimizing Procurement of Goods for a Food-based FMCG Company

4.3 Festivals vs Sales Correlation Analysis

Analysis Process
For the purpose of ascertaining whether festivals and sales have a positive correlation, the following
steps were followed, including the pre-processing of the dataset.

• Step 1: A list of all the festivals in India in the FY 2020-2022 was created. Online sources
were used for this purpose.

• Step 2: The data was in the form of individual bills with many repeating dates, so a pivot
table in MS Excel was created to get individual dates.

• Step 3: A list of all the dates beginning from April 2020 to March 2022 was created in MS
Excel.

• Step 4: Using VLOOKUP of MS Excel, if there existed a Sale on the corresponding date, ”1”
was appended to the cell, else ”0” was appended. Same was done for every festival. The data
was grouped based on dates. Three different datasets were created - Daily, Fortnightly and
Monthly.

• Step 5: The data created above was saved in the CSV format to perform further analysis.

• Step 6: A correlation analysis was done using Pandas Python Library among the features whose
code is shown in Figure 4.

• Step 7: Using MS Excel, a pivot table was created that grouped the data into months and plotted
the trend line of Total Amount and Frequency of Festival each month as shown in Figure 6.

Reasoning
To determine whether a sale occurs during a festival season, it is only necessary to ascertain
whether a sale was made on a specific day and not the actual amount of the sale. Correlation
analysis using Pandas Library can indicate whether any significant orders were placed on the day of
the festival. Additionally, monthly analysis performed using MS Excel can identify whether any
significant orders were placed during the same month as the festival. In the event of a positive
correlation between festivals and sales, it is advisable to ensure that the inventory has a higher
number of units available to meet the anticipated demand during the festival season. This helps to
ensure that orders can be fulfilled promptly, and customers can receive their orders within a
reasonable time frame.

Figure 4: Python Code to find correlation between festivals and sales

Page 7 of 9
 Optimizing Procurement of Goods for a Food-based FMCG Company

4.4 Seasons vs Sales Correlation Analysis

Analysis Process
For the purpose of ascertaining whether seasons and sales have a positive correlation, the following
steps were followed, including the pre-processing of the dataset.

• Step 1: The data was in the form of individual bills containing SKUs with many repeating
dates, so a pivot table in MS Excel was created to get unique dates.

• Step 2: The data was grouped into months, and a Line chart was plotted for Months vs
Quantity of SKUs Sold.

Reasoning
Since the seasons last for 3-4 months, a monthly analysis would be sufficient to find a seasonal
trend. In this analysis, the actual number of SKU units sold is important as it will also capture the
volume of the order placed in a particular season.

5 Results and Findings

5.1 SKU Pareto Analysis
It is clear from the analysis shown in Figure 3, that the Top SKUs that cover greater than 85% of
sales are:

1. SKU6 -

2. SKU12 -

3. SKU2 -

4. SKU10 -

5. SKU17 -

Upon further discussions with the business representative, this result was not considered surprising,
since these 5 SKUs were the oldest ones introduced by the company. Therefore only these SKUs
should be kept in the inventory.

5.2 Festivals vs Sales Correlation Analysis

From Figure 5, it is clear that there is no Daily, Fortnightly or Monthly correlation between festivals
and sales. This result can also be confirmed by the trend line plotted in Figure 6. It is safe to
conclude that our original hypothesis, that the sales are affected by the festivals is rejected, and
cannot be used for effective inventory keeping.

Figure 5: Correlation Heatmap between Festivals and Sales

Page 8 of 9
 Optimizing Procurement of Goods for a Food-based FMCG Company

Figure 6: Trend line between Festivals and Sales

5.3 Seasons vs Sales Correlation Analysis

From Figure 7, it is clear that the sales increase in the winter months of December, January and
February as well as in the months of June, July and August thereby confirming our original
hypothesis that the sales are seasonal in nature. We can make a model that predicts how much
inventory to keep of individual SKU based on this plot.

Figure 7: Monthly Number of SKUs sold for the Top 5 SKUs

6 Conclusion and What next?

The original hypothesis in the project proposal that the sales are affected by the festivals stands re-
jected. The second hypothesis that the sales are affected by the seasons of the year has been
confirmed as there is a clear seasonal trend in the sales of the company.

The seasonal trends can be used to make a predictive model using machine learning that will predict
the number of units of SKUs to keep in the inventory. This can be used as an effective inventory
management model. As stated earlier, the company wants to optimize the procurement process by
effective inventory management. If the number of SKUs required by the company are known on a
particular date because of effective inventory management, then the gap between procurement of
the raw material and delivery of end products may be optimally reduced, thereby solving our
original problem of Optimizing the Procurement of goods for a Food Based FMCG company.

Page 9 of 9