Supply_Chain_Management_EDA

Contents

• Introduction
• Data Exploration
• Data Preprocessing
• Data Visualization
• Suggestions

Introduction

A Fast Moving Consumer Goods (FMCG) company entered into the instant noodles business two years back. Their higher management has noticed that there is a mismatch in the demand and supply. Where the demand is high, supply is pretty low and vice-versa which results in a loss in inventory cost and ultimately loss to the company. Hence, the higher management wants to analyze the supply quantity in each and every warehouse in the entire country.

Data Exploration

Data Dictionary

• Ware_house_ID : Unique Warehouse id where product is prepared for dispatch
• WH_Manager_ID : Manager Id present in the warehouse
• zone : Zone of the Warehouse
• WH_regional_zone : Regional Zone of the warehouse
• num_refill_req_l3m : Refilling request received by the warehouse in the last 3 months
• transport_issue_l1y : No. of transport issued for warehouse in last 1 year
• Competitor_in_mkt : No. of competitors in the market
• retail_shop_num : Number of retail shops who sell noodlesproduced by the warehouse
• wh_owner_type : The warehouse is owned by the company or it is on rent
• distributor_num : No. of distributor who works between warehouse and retail shops
• flood_impacted : Is the warehouse in a flood impacted area or not
• flood_proof : Warehouse is having flood proof indicator
• electric_supply : Does the warehouse have proper electric supply along with some power backup
• dist_from_hub : distance from the warehouse to production hub
• workers_num : no. workers in the warehouse
• wh_est_year : warehouse establishment year
• storage_issue_reported_l3m : storage issues reported by the warehouse in the last 3 months
• govt_check_l3m : Government checking in last 3 months
• temp_reg_mach : warehouse having temperature regulating machine indicator or not
• approved_wh_govt_certificate : Type of approval warehouse having been issued by government
• wh_breakdown_l3m : Number of times the warehouse faces thebreakdown in the last 3 months
• product_wg_ton : Product weight

• Dataset has 25 columns and 16620 rows.
• No. of columns with int datatype : 15
• No. of columns with object datatype : 8
• No. of columns with float datatype : 2

• No. of unique values in each column :
  Unnamed: 0 16620
  WH_Manager_ID 16620
  Ware_house_ID 16620
  retail_shop_num 4356
  product_wg_ton 4248
  dist_from_hub 217
  workers_num 60
  distributor_num 56
  storage_issue_reported_l3m 37
  govt_check_l3m 32
  wh_est_year 28
  Competitor_in_mkt 12
  num_refill_req_l3m 9
  wh_breakdown_l3m 7
  transport_issue_l1y 6
  WH_regional_zone 6
  approved_wh_govt_certificate 5
  zone 4
  WH_capacity_size 3
  flood_proof 2
  flood_impacted 2
  temp_reg_mach 2
  Location_type 2
  wh_owner_type 2
  electric_supply 2

Data Preprocessing

This project mainly focus on analyzing impact of features provided in the dataset on target variable product weight.

1. Drop irrelevent columns

1. Unnamed: 0
2. Ware_house_ID
3. WH_Manager_ID

These three columns each contain entirely distinct values. In the context of this project, the uniqueness of these columns holds no relevance, as they are only analyzing product quantity. Therefore, these three columns have been removed.

2. Missing values

Among the 25 columns, null values are present in three columns. These columns correspond to the number of workers in the warehouse (workers_num), the year of warehouse establishment (wh_est_year), and the grade of government certificate approved for the warehouse (approved_wh_govt_certificate).

Plot of null values

% of null values in eachg column:

workers_num(4.01 %) wh_est_year(47.29 %) approved_wh_govt_certificate(3.6 %)

workers_num :

• numerical column
• positively skewed Filled null values with median.

wh_est_year :

Since half of the values in the column wh_est_year are null, decided to drop it.

approved_wh_govt_certificate:

• categorical column Filled null values with mode.

3. Duplicate values

No dupliocate values were present in the dataset.

4. Outliers

Outliers were present in some of the columns. plotted boxplot to locate outliers.

Determined the count of outliers in every column. Among them, the 'flood_proof' and 'flood_impacted' columns, each containing only two distinct values, exhibit significant disparities in their value distributions and lack correlation. one of the unique values is considered as outlier due to its significantly lower proportion. Consequently, both of these columns were removed from the dataset.

Outliers were detected and removed

Data Visualization

Relationship between other features and target variable Product weight

• Warehouse location type & Owner Type

• Electric_supply & temp_reg_mach availability

• Storage issue reported in the last 3 months

• No. of warehouse breakdown in the last 3 months

• zone & Regional_zone

• Type of approval by government

• govt checking in last 3 months

• No. of workers

-Warehouse capacity size & Transport issue

• Competitor_in_mkt

• No. of Distributors

• No. of refill request in last 3 months

• Distance from hub

• Retail shop number

• Correlation Heatmap

Suggestions

• Focus on expanding warehouse operations in rural areas to capitalize on the higher average product weights and the larger share of total product weight, which can potentially lead to increased overall productivity and profitability.
• Prioritize investment in temperature regulating machines and enhancing electric supply availability for warehouses to improve product weight and overall warehouse performance.
• It is evident that, as production increases, storage issue increases. Hence, storage capacity must be upgraded to accomodate higher product weights. This highlights the importance of regularly assessing and expanding storage infrastructure in line with the production increments to ensure smooth warehouse operations and prevent bottlenecks.
• Implement proactive maintanance protocols. Invest in scalable solutions to accomodate higher production levels while minimizing the risk of operational disruptions caused by breakdowns. Conduct comprehensive analysis to identify root causes of braekdowns during periods of heightened production.
• Explore quality standards associated with different government approval grades. Identify opportunities to enhance product quality standards across all warehouse facilities.
• Workers number does not significantly affect production. Examine various factors such as technology adoption, training programs, and workflow optimization strategies to understand their collective impact on production efficiency.