Analyzing On-Time Delivery for E-
Commerce Logistics
MPBAG507-Programming for Analytics Project
Presented by:
Ayushman Pant - 2024H1540803P
Abhijit Mishra - 2024H1540825P
Rajdeep Nath - 2024H1540830P
Sameer Ravi Kanth - 2024H1540835P
Overview
Meeting customer demands on-time is one of the top priorities
of e-commerce companies. The dataset in this case is based on
product shipment tracking by an international e-commerce
company that wants key insights from its customer database.
Through data analytics, it will be possible to gain useful
insights about what on-time shipping logistics. Through them,
various kinds of analyses have been done, including predictive
analytics
Goal
Analyze customer shipment data to understand delivery
patterns and factors influencing timely delivery.
Dataset Details
• 10,999 records with 12 features.
• Key attributes include shipping methods, customer ratings, product importance, and delivery status.
Data
Exploration
Data Cleaning & Describing:
• First, dataset was cleaned & described with appropriate functions
commonly used in Python programming.
• Cleaning was done on the conditions of checking for null & missing
values.
• Descriptions are critical for checking possible unique values as well
as for describing all the features of attributes.
Data Exploration
contd.
Distributions:
• Histograms plotted for numerical features (e.g.,
Cost_of_the_Product, Weight_in_gms).
• Insights into the spread and central tendency of data are done
with univariate analysis.
Data Exploration
contd.
Categorical Analysis:
• Count plots for warehouse, shipment modes, customer
rating & whether shipments are reaching in time or not.
• The comparison of values & categories underscore where
exactly the conditions lie again through univariate
analysis.
Relationship between
Variables
• Bivariate analysis is performed in order to find relationship
between two variables.
• Categorical vs Target: Count plots showing delivery status
([Link].Time_Y.N) across categorical features (e.g.,
Warehouse_block and Gender).
Relationship between
Variables contd.
• Numerical vs Target: Box plots visualizing how
numerical features like Discount_offered and
Weight_in_gms influence delivery status.
Advanced
Visualization Pairwise Relationships: Pair plots
exploring interactions between numerical
features with [Link].Time_Y.N as the
hue.
Advanced
Visualization
contd.
• Heatmap highlighting correlations between features.
• Bar plot showing the strength of relationships with the target
variable.
Advanced Insights into
Predictive Analytics
The Confusion Matrix provides the values that are
predicted against those that are actual in terms of
either being on time or not on time.
Factors significantly correlated with on-time
delivery:
• Discounts offered.
• Prior purchases and product weight.
Categorical influences:
• Shipment mode and warehouse block.
Conclusion
• The significance of shipping being on time highlights a major concern for
e-commerce companies.
• This is underlined by the factors showing different levels at different
unique values.
• However, the target variable of whether shipping has reached on time
reflects immense variations & differences encountered in this case.
• Through univariate & bivariate analysis, these underscored the crucial
perspectives on which the companies must focus on.
• However, advanced visualization through correlation analysis & confusion
matrix offers insights about how discounts, prior purchases & product
weight are the ones most correlated to the output.
• Shipment mode & warehouse block- are also correlated as categorical
variables to this output.
Thank You