Customer Analytics Turtle Games

Overview

This project was completed as part of the Data Analytics Career Accelerator from The London School of Economics and Political Science (LSE) in collaboration with FourthRev.

Objective
To identify customer trends that can help Turtle Games improve its overall sales performance through data-driven insights and customer segmentation.

Business Problem

"How can Turtle Games use customer trends to improve overall sales performance?"

The analysis focused on:

Understanding the relationship between customer demographics (age, remuneration, spending score) and loyalty points.
Exploring customer sentiment from reviews and summaries to gauge perception.
Segmenting customers based on income and spending patterns to tailor marketing strategies.

Tools & Technologies

Category	Tools Used
Data Analysis	Python (pandas, numpy, scikit-learn), R (dplyr)
Data Visualisation	Python (Matplotlib, Seaborn), R (ggplot2, ggcorrplot)
Statistical Analysis	Python (Scipy, Statsmodels), R (moments)
Natural Language Processing (NLP)	Python (nltk, wordcloud, textblob, collections, vaderSentiment)
Environment	Jupyter Notebook, RStudio

Analytical Approach

1. Exploratory Data Analysis (EDA)

Assessed distributions for key variables including age, remuneration, spending score, and loyalty points.
Calculated skewness and kurtosis metrics to evaluate structural data symmetry and tail behaviour.

2. Correlation Analysis

Generated an interactive correlation heatmap to map linear dependencies across variables.
Identified strong positive correlations linking spending score and remuneration with loyalty points.

3. Customer Segmentation

Partitioned the customer base into discrete behavioural categories: Low, Medium, and High income/spending bands.
Analysed variations in loyalty point accumulation profiles across each distinct group.

4. Sentiment Analysis

Evaluated customer review and summary text data to calculate baseline sentiment polarity metrics.
Generated text word clouds mapped against polarity scores to isolate high-frequency terms and determine overall perception tone.

Key Insights

Demographic Overview

Average customer age: 39 years (range: 17–72).
Average annual remuneration: £48,080 (range: £12,300–£112,340).
Average spending score: 50 (range: 1–99).
Average loyalty points: 1,578 (range: 25–6,847).

Distribution Analysis

Variable	Skewness	Kurtosis	Interpretation
Age	+0.61	2.80	Slight right skew, light-tailed
Remuneration	+0.41	2.59	Right skew, light-tailed
Spending Score	-0.04	2.11	Nearly symmetric
Loyalty Points	+1.46	4.70	Heavily right skewed, outlier-sensitive

Correlation Findings

Correlation Heatmap

Teal is positive, ochre negative; strength shown by depth

Remuneration ↔ Loyalty Points: Moderate positive correlation (r = 0.62)
Spending Score ↔ Loyalty Points: Stronger positive correlation (r = 0.67)
Indicates that higher spenders and earners accumulate more loyalty points.

Customer Segmentation Insights

Loyalty Points by Remuneration Group

Group	Mean Loyalty Points	Median Loyalty Points
Low Earners	725	724
Medium Earners	1,551	1,463
High Earners	2,503	2,262

Higher earners accumulate more loyalty points consistently.

Loyalty Points by Spending Group

High Spending customers accumulate the most loyalty points.
Medium-High group shows balanced and consistent loyalty accumulation.
Low Spending group occasionally contains outliers with very high loyalty points.

Customer Clustering

Identified five customer clusters based on spending and remuneration patterns.
These clusters provide opportunities for targeted promotions and personalized campaigns.

Sentiment Analysis Results

Summary Sentiment

Mean polarity: +0.22 → Generally positive tone
Common positive words: "great", "good", "love", "cute"
Occasional neutral/negative mentions: "game" (some critiques)

Review Sentiment

Mean polarity: +0.217 → Slightly positive overall sentiment
Frequent words: "great", "fun", "love", "good"
Suggests overall positive customer experience with some minor product feedback.

Recommendation

Problem question	Actionable Insights	Business Recommendation
How can Turtle games use the customer trends to improve overall sales performance?	There was a positive correlation between remuneration and spending score with loyalty points, suggesting that higher earning customers and higher spending customers tend to accumulate more loyalty points. Additionally, grouping the data by spending score revealed that customers with higher spending scores also had higher loyalty points, confirming that spending behaviour is a key factor in loyalty point accumulation.	In order to improve overall sales performance, Turtle games should provide benefits to the customer for accumulating loyalty points. For instance, customers could get different levels of discount (5%, 10%, 15%, 20%) for purchasing products in exchange for the loyalty points. In this way, customers see benefit in accumulating loyalty points and since accumulating loyalty points is directly correlated with customer spending, Turtle games can improve overall sales performance by incentivising the customers to accumulate more loyalty points.

Porficiencies Demonstrated

Data Wrangling & Descriptive Statistics: Programmatically cleaning raw datasets, identifying outliers, and generating core summary statistics (mean, median, skewness, kurtosis).
Correlation & Clustering Analysis: Executing paired correlation tests to isolate dependencies and executing customer segmentation to isolate high-value behavioral groups.
Natural Language Processing (NLP): Utilizing lexicon-based classifiers (TextBlob, vaderSentiment) to quantify text polarity and extract user perception trends.
Data Visualization & Interpretation: Constructing scannable exploratory heatmaps and word clouds to translate abstract data tables into strategic business realities.
Structured Professional Communication: Synthesizing dense statistical results into clean, actionable executive summaries tailored for non-technical stakeholders.

Summary

This project demonstrates how data analytics can uncover key customer trends that drive actionable sales strategies.
Through correlation, segmentation, and sentiment analysis, the study provides evidence-based recommendations for improving Turtle Games' sales and loyalty performance.