top of page

Exploratory Data Analysis of Customer Support Tickets

Data analysis of customer support tickets for various tech products, related to hardware issues, software bugs and network problems, as well as information about the customer, ticket type, priority and other relevant details.

Business question: What products trigger more tickets, and how is it influencing customer satisfaction?

Tools used: Python.

Proj-Support-Tickets_2x.jpg
Key Insights

  • The top 5 products that generated more tickets were Canon EOS, IPhone, Canon DSLR Camera, GoPro Hero and Microsoft Office.

  • The top 5 subjects that generated more tickets were Network problem, Software bug, Product compatibility, Product recommendation and Product Setup.

  • Ticket priority is homogeneously distributed ranging from 23% Low to 26% Critical.

  • Customer Satisfaction Rating is also homogeneously distributed ranging from 20% Level 1 - Low satisfaction to 19%  Level 5 - High satisfaction. 

  • Ticket type: Refund and Cancellation requests combined account for 41%, a strong signal of low customer satisfaction.

  • Channels used for customer support have little variation from 26% of emails to 24% of chat, its customers are spread across the different ways of communication, it may indicate different ages, backgrounds and education levels.

  • It is not clear to set a correlation between Ticket Priority and Customer Satisfaction.

  • The same happens between Ticket Type and Customer Satisfaction, this may be due to data homogeneity.

1. Data Source

Data source from Kaggle.
The dataset consists of customer inquiries related to hardware issues, software bugs, network problems, account access, data loss, and other support topics.

It provides information about the customer, the product purchased, the ticket type, the ticket channel, the ticket status, customer satisfaction and other relevant details.
CSV file with 8469 rows and 17 columns.

2. Data Exploration and Visualization

In Python, Exploratory Data Analysis (EDA) organized in 11 steps:
 

  • Step 1: Introduction

  • Step 2: Import Libraries

  • Step 3: Reading Dataset

    • Check for Unique and Duplicated Values

    • Missing Values

  • Step 4: Data Reduction

  • Step 5: Feature Engineering

    •  Creating Features

  • Step 6: Data Cleaning / Wrangling

  • Step 7: Statistics Summary

  • Step 8: EDA Univariate Analysis

  • Step 9: Data Transformation

    • Standardization of numerical variables (Z-Score Normalization, using StandardScaler)​

  • Step 10: EDA Bivariate Analysis

    • Convert categorical columns to numerical ones, for comparision (using factorize)​

  • Step 11: EDA Multivariate Analysis

​

In terms of Visualization, use Seaborn and Matplotlib libraries to create several graphs:
 

  • Pie chart to get the distribuition by gender, resolution, ticket priority, customer satisfaction rating.

  • Histogram for customer age, customer satisfaction rating by ticket priority, customer satisfaction rating by ticket type.

  • Count plot (vertical and horizontal) to get the sum of product purchase, brand, ticket subject, ticket channel, ticket type.

  • Heatmap for correlation of all variables.

Other projects

bottom of page