Datasets:

MatanKriel
/

Matan-Kriel-Customer-Churn-dataset

Modalities:

Tabular

Text

Formats:

Size:

Libraries:

Dataset card Data Studio Files Files and versions

xet

Community

Dataset Viewer

Auto-converted to Parquet Duplicate

Split (1)

train · 441k rows

CustomerID float64 2 450k ⌀	Age float64 18 65 ⌀	Gender stringclasses 2 values	Tenure float64 1 60 ⌀	Usage Frequency float64 1 30 ⌀	Support Calls float64 0 10 ⌀	Payment Delay float64 0 30 ⌀	Subscription Type stringclasses 3 values	Contract Length stringclasses 3 values	Total Spend float64 100 1k ⌀	Last Interaction float64 1 30 ⌀	Churn float64 0 1 ⌀
2	30	Female	39	14	5	18	Standard	Annual	932	17	1
3	65	Female	49	1	10	8	Basic	Monthly	557	6	1
4	55	Female	14	4	6	18	Basic	Quarterly	185	3	1
5	58	Male	38	21	7	7	Standard	Monthly	396	29	1
6	23	Male	32	20	5	8	Basic	Monthly	617	20	1
8	51	Male	33	25	9	26	Premium	Annual	129	8	1
9	58	Female	49	12	3	16	Standard	Quarterly	821	24	1
10	55	Female	37	8	4	15	Premium	Annual	445	30	1
11	39	Male	12	5	7	4	Standard	Quarterly	969	13	1
12	64	Female	3	25	2	11	Standard	Quarterly	415	29	1
13	29	Male	18	9	0	30	Premium	Quarterly	930	18	1
14	52	Female	21	6	3	26	Premium	Monthly	830	19	1
15	22	Male	41	17	10	25	Basic	Quarterly	265	23	1
16	48	Female	35	25	1	13	Basic	Annual	518	17	1
17	24	Male	4	9	4	22	Standard	Quarterly	204	4	1
18	49	Male	56	17	2	30	Standard	Quarterly	975	17	1
19	19	Female	38	23	7	11	Basic	Quarterly	978	3	1
20	47	Male	41	14	1	5	Premium	Annual	151	19	1
21	24	Male	44	13	5	4	Premium	Monthly	669	13	1
22	42	Male	15	16	2	14	Premium	Quarterly	262	16	1
23	57	Female	55	27	3	3	Basic	Annual	212	10	1
24	39	Female	43	2	4	15	Basic	Monthly	577	6	1
25	27	Male	44	28	8	18	Standard	Annual	436	30	1
26	27	Female	52	8	7	3	Standard	Monthly	434	19	1
27	59	Male	26	21	0	10	Premium	Monthly	822	17	1
28	21	Male	2	21	7	22	Basic	Annual	435	21	1
29	60	Female	18	16	8	28	Basic	Quarterly	830	22	1
30	65	Female	29	29	0	5	Premium	Annual	857	18	1
31	35	Female	38	20	6	2	Premium	Annual	574	19	1
32	18	Male	37	15	8	6	Premium	Monthly	800	29	1
33	56	Female	59	5	10	27	Standard	Quarterly	424	2	1
35	35	Male	40	27	8	28	Premium	Monthly	232	17	1
36	29	Male	43	12	2	10	Premium	Annual	289	6	1
37	20	Male	37	24	7	5	Standard	Annual	371	20	1
38	63	Female	51	3	5	12	Standard	Annual	288	15	1
39	22	Female	39	8	2	4	Standard	Annual	939	21	1
40	25	Female	53	2	10	13	Basic	Quarterly	336	13	1
42	28	Male	24	24	8	6	Standard	Monthly	572	28	1
43	51	Male	30	4	10	29	Premium	Monthly	770	23	1
44	32	Male	6	22	3	12	Basic	Monthly	413	20	1
45	38	Female	28	23	6	17	Basic	Quarterly	993	28	1
46	52	Male	17	26	4	6	Basic	Monthly	871	26	1
48	37	Male	30	30	7	13	Basic	Quarterly	929	13	1
49	31	Female	4	29	7	22	Standard	Annual	488	6	1
50	30	Male	24	7	6	13	Standard	Annual	874	3	1
51	23	Female	26	21	7	24	Basic	Quarterly	988	20	1
52	35	Male	37	11	7	13	Standard	Monthly	631	7	1
53	21	Female	56	11	9	9	Basic	Quarterly	626	1	1
54	56	Female	44	11	2	11	Standard	Annual	583	9	1
55	53	Female	18	21	1	23	Basic	Quarterly	717	15	1
56	41	Female	18	27	10	3	Basic	Monthly	111	19	1
57	33	Male	6	12	7	29	Standard	Quarterly	869	10	1
58	18	Male	55	29	9	21	Standard	Quarterly	296	22	1
59	26	Male	60	9	10	1	Standard	Monthly	643	2	1
60	42	Female	7	9	1	11	Basic	Annual	196	18	1
61	36	Male	2	15	3	29	Premium	Monthly	501	4	1
62	22	Female	43	16	8	21	Standard	Monthly	966	30	1
63	47	Male	59	24	6	26	Premium	Annual	288	26	1
64	30	Male	52	23	1	27	Basic	Annual	400	20	1
66	21	Female	51	1	8	6	Premium	Annual	618	13	1
67	27	Female	40	17	3	1	Basic	Annual	999	17	1
68	27	Male	34	13	4	26	Basic	Quarterly	214	6	1
69	32	Male	3	5	1	6	Standard	Annual	282	20	1
70	44	Female	43	1	8	28	Standard	Monthly	548	28	1
71	38	Male	10	9	6	5	Premium	Annual	820	18	1
72	34	Female	5	18	3	29	Premium	Quarterly	575	10	1
73	42	Male	29	23	0	6	Premium	Monthly	736	5	1
74	61	Male	41	4	7	23	Basic	Quarterly	753	23	1
75	22	Male	45	13	5	16	Basic	Quarterly	538	26	1
76	59	Female	53	13	0	4	Standard	Quarterly	688	14	1
77	53	Male	54	18	10	0	Basic	Monthly	679	1	1
78	37	Male	17	23	9	23	Premium	Quarterly	530	19	1
79	65	Female	33	28	9	22	Premium	Monthly	383	17	1
80	61	Female	6	7	4	26	Standard	Annual	369	20	1
81	40	Female	58	8	2	10	Basic	Monthly	760	28	1
82	63	Female	38	19	7	21	Standard	Quarterly	768	7	1
83	19	Male	53	3	6	17	Standard	Monthly	808	19	1
84	45	Female	25	4	8	13	Premium	Quarterly	681	26	1
85	59	Male	17	21	1	24	Standard	Quarterly	858	2	1
86	55	Female	54	2	5	0	Standard	Monthly	159	29	1
87	44	Male	13	12	7	7	Standard	Monthly	624	15	1
88	61	Male	7	12	7	2	Premium	Annual	925	9	1
89	46	Male	41	12	3	25	Premium	Annual	531	11	1
90	22	Male	45	17	3	28	Basic	Quarterly	769	3	1
92	46	Female	47	4	5	10	Standard	Annual	299	5	1
93	64	Male	31	20	9	24	Premium	Quarterly	785	16	1
94	28	Female	22	16	1	7	Standard	Monthly	389	16	1
95	35	Female	19	28	4	28	Basic	Quarterly	507	5	1
96	55	Female	34	6	2	7	Basic	Annual	392	5	1
97	35	Male	3	17	6	23	Basic	Annual	392	7	1
98	45	Female	23	4	5	1	Premium	Quarterly	699	12	1
99	24	Male	1	20	0	17	Basic	Quarterly	268	4	1
100	44	Female	8	11	7	26	Premium	Monthly	139	3	1
102	60	Male	55	7	3	18	Standard	Monthly	871	4	1
103	40	Female	46	24	3	1	Basic	Monthly	404	28	1
104	26	Male	15	11	0	20	Basic	Annual	269	1	1
105	58	Female	28	17	0	24	Premium	Quarterly	820	29	1
106	56	Female	16	11	5	18	Premium	Monthly	590	27	1
107	42	Female	12	6	8	25	Basic	Monthly	566	11	1
108	52	Male	29	2	2	4	Basic	Annual	435	27	1

End of preview. Expand in Data Studio

YAML Metadata Warning: empty or missing yaml metadata in repo card (https://huggingface.co/docs/hub/datasets-cards)

Project Overview:

Uncovering Critical Churn Drivers

This project employed Exploratory Data Analysis (EDA) techniques on a comprehensive customer dataset with the primary objective of diagnosing the root causes of severe customer attrition.

The Business Challenge: The initial analysis revealed an alarmingly high baseline churn rate of 56.7%, indicating a massive challenge in customer retention and significant revenue leakage. The core mandate was to move beyond surface-level assumptions and utilize the data to identify which customers were leaving, and why.

Data Preparation & Cleaning

Missing Data Handling: Verified and removed the single row with entirely missing values.
Data Deduplication: Checked the dataset for and removed any exact duplicate customer entries to prevent analytical bias.
Converted key numerical count features (e.g., Tenure, Support Calls) to memory-efficient integer types (int64).
Categorical Consistency: Performed a value-count check on all object columns (Gender, Contract Length, etc.) and confirmed they were free of typos, mixed casing, or whitespace errors.
Outlier Detection: Used descriptive statistics (.describe()) and Box Plots to visually inspect all continuous numerical features. No extreme or non-plausible outliers were detected, justifying the decision to keep all observed data points.

Exploratory Data Analysis (EDA)

🔹 Research Question 1:

Does Contract Length affect Churn Rate?

Analysis of Contract Length vs. Churn Rate confirms that our hypothesis was only partially correct.

Key Findings: Monthly Contract Risk: The Monthly contract length carries the highest churn rate by a significant margin. This aligns with the expectation that customers with shorter commitments are the least loyal.

Annual vs. Quarterly Parity: There is no significant difference in the churn rate between customers on Quarterly and Annual plans. Both long-term contract types demonstrate a much lower risk of churn compared to the Monthly plan.

Business Conclusion and Strategy: This finding has a major business implication for sales and retention strategy.

Instead of solely focusing sales efforts on the Annual plan, which may be harder to close due to the longer commitment, the business can confidently promote the Quarterly plan. By shifting focus to the Quarterly option, the company achieves the same low churn risk as the Annual plan while potentially increasing the volume of long-term contract sales.

🔹 Research Question 2:

How is Total Spend related to the likelihood of Churn?

Analysis of the distribution of Total Spend by churn status reveals a highly counter-intuitive and critical business finding:

High Spenders Churn: The vast majority of customers who Churned (1) are concentrated in the higher Total Spend bracket (roughly 500𝑡𝑜 1,000).

Low Spenders Stay: Conversely, the customers who Did Not Churn (0) are overwhelmingly concentrated in the lower Total Spend bracket (roughly 100𝑡𝑜 550).

Business Implication: Contrary to the common assumption that low-value customers are the biggest churn risk, this company is primarily losing its high-value customers. This suggests that customer dissatisfaction is not driven by the cost of the service, but rather by issues like poor service quality, inadequate support, or unmet expectations that are particularly felt by their most valuable customer segment.

This finding immediately shifts the focus of the retention strategy from price sensitivity to improving the customer experience for premium users.

🔹 Research Question 3:

Is there a connection between the number of Support Calls a customer makes and their likelihood of Churning?

The visualization confirms a clear and critical trend: there is a strong positive correlation between the number of Support Calls a customer makes and their likelihood of Churning.

Key Findings: Direct Relationship: As the volume of support calls increases, the average churn rate consistently rises. This suggests that customers are reaching out because they are experiencing unresolved frustration or ongoing service issues.

The Critical Threshold: The data reveals a definitive threshold: customers who make six or more Support Calls have a 100% Churn Rate. This group is effectively guaranteed to leave the company.

Business Conclusion and Strategic Action: This finding provides the business with an immediate, actionable strategy to improve retention:

The "Sweet Spot" for Intervention: The moment a customer reaches four or five Support Calls, they are in the high-risk zone but still potentially savable. Before they hit the critical threshold of six calls, the company has its last, best chance to intervene.

Key Insights

High-Value Churn: The company is losing its highest-spending customers (Total Spend $500-$1,000) due to experience failures, not price.
Support Call Red Line: Customers who reach six or more Support Calls are guaranteed to churn (100% loss). Intervene urgently at 4-5 calls.
Contract Parity: Quarterly contracts deliver the same low churn rate as Annual contracts. Promote Quarterly for easier long-term sign-ups.

Tools & Libraries

Python
Pandas
NumPy
Matplotlib
Seaborn

Author

Name: Matan Kriel

Youtube URL for the Video review of the project:

URL: https://youtu.be/CAZv5PS4P1A

license: mit

Downloads last month: 7

Size of downloaded dataset files:

30.5 MB

Size of the auto-converted Parquet files:

8.04 MB

Number of rows:

440,833