Cluster Analysis of Activa Scooty

Authors:

1. Tanay Harsh (021330624078)
2. Krish Chauhan (02133024003)
3. Mohit Kumar Sah (02133024340)

 

 

Case Processing Summarya,b
Cases
Valid		Missing		Total
N	Percent	N	Percent	N	Percent
54	100.0	0	.0	54	100.0
a.  Squared Euclidean Distance used
b. Average Linkage (Between Groups)

 

Introduction

In the context of consumer preferences and segmentation, understanding the extent to which various product characteristics differentiate user clusters can provide valuable insights for targeted marketing and product optimization. This analysis explores the clustering of seven key attributes related to a product’s functionality and user appeal: efficiency, comfort, storage, combi-braking system (CBS), suspension, affordability, and illumination. By examining variance in these characteristics across clusters, we aim to identify whether these features contribute meaningfully to segmenting consumers with differing preferences or needs.

Clustering analysis, followed by ANOVA and effect size measurement, allows us to assess whether differences across clusters are statistically significant and practically meaningful. This approach highlights whether clusters reflect genuine distinctions in user preferences for specific characteristics or if observed differences are likely due to chance. Effect size metrics, including Eta-squared, Epsilon-squared, and Omega-squared, further clarify the magnitude of clustering effects, adjusting for biases and quantifying how much of the variation in each characteristic is explained by clustering.

Objective

The primary objective of this analysis is to determine the extent to which selected product characteristics (efficiency, comfort, storage, CBS, suspension, affordability, and illumination) differentiate clusters. Specifically, the analysis aims to:

1. Assess Significance: Evaluate whether variance in each characteristic across clusters is statistically significant using ANOVA.
2. Measure Effect Size: Use Eta-squared, Epsilon-squared, and Omega-squared to quantify the practical significance of clustering effects on each characteristic, with an emphasis on understanding the degree of variance attributed to cluster differences.
3. Identify Patterns: Interpret findings to identify characteristics with potential clustering relevance and suggest whether additional variables or methods might enhance cluster differentiation.

The results will inform whether these characteristics drive meaningful distinctions across clusters, providing a foundation for potential segmentation strategies and identifying areas for further investigation in understanding user preferences.

 

 

Agglomeration Schedule
Stage	Cluster Combined		Coefficients	Stage Cluster First Appears		Next Stage
	Cluster 1	Cluster 2		Cluster 1	Cluster 2
1	1	3	8.000	0	0	6
2	32	53	32.000	0	0	10
3	12	29	32.000	0	0	23
4	11	47	36.000	0	0	17
5	35	45	36.000	0	0	18
6	1	40	38.000	1	0	21
7	9	48	40.000	0	0	25
8	5	44	42.000	0	0	16
9	43	49	46.000	0	0	29
10	17	32	48.000	0	2	18
11	28	46	52.000	0	0	29
12	19	39	55.000	0	0	24
13	7	41	57.000	0	0	23
14	15	26	59.000	0	0	31
15	4	18	60.000	0	0	39
16	5	6	61.000	8s	0	33
17	11	21	63.000	4	0	31
18	17	35	63.333	10	5	22
19	22	36	64.000	0	0	34
20	13	16	67.000	0	0	34
21	1	33	69.000	6	0	35
22	17	27	70.600	18	0	38
23	7	12	71.000	13	3	36
24	19	54	74.500	12	0	37
25	9	14	77.000	7	0	28
26	37	38	80.000	0	0	40
27	30	42	82.000	0	0	44
28	9	24	84.000	25	0	38
29	28	43	85.000	11	9	40
30	20	31	85.000	0	0	45
31	11	15	86.167	17	14	33
32	23	50	89.000	0	0	39
33	5	11	92.733	16	31	37
34	13	22	94.000	20	19	44
35	1	52	95.750	21	0	36
36	1	7	104.550	35	23	43
37	5	19	108.625	33	24	42
38	9	17	117.167	28	22	42
39	4	23	119.500	15	32	48
40	28	37	120.750	29	26	49
41	25	34	125.000	0	0	52
42	5	9	127.555	37	38	46
43	1	8	127.889	36	0	45
44	13	30	132.000	34	27	50
45	1	20	132.500	43	30	46
46	1	5	148.317	45	42	47
47	1	10	156.939	46	0	48
48	1	4	163.456	47	39	49
49	1	28	166.439	48	40	50
50	1	13	182.811	49	44	51
51	1	51	201.780	50	0	52
52	1	25	216.598	51	41	53
53	1	2	242.981	52	0	0

 

 

 

 

 Case Processing Summary Interpretation

 

The Case Processing Summary table tells us about the completeness of the dataset and the methods applied in the clustering.

 

1. Dataset Completeness:

   – Valid Cases: All 54 cases in this dataset are complete, with no missing entries, giving us a full dataset (100% validity) for analysis. This ensures that all data points contribute to forming clusters, improving the stability and accuracy of the resulting groups.

   – Implication: A complete dataset increases the reliability of clustering results. If cases were missing, it could skew the clusters or make some groupings less representative of underlying patterns.

 

2. Distance Metric:

   – Squared Euclidean Distance: This metric measures similarity by calculating the squared differences between points. Squaring the differences means that larger discrepancies between data points are given more weight, which helps highlight significant distinctions in the dataset.

   – Implication: Using this metric emphasizes outliers or larger differences, helping to form clusters where cases within each group are relatively similar, and distinct from cases in other clusters.

 

3. Linkage Method:

   – Average Linkage (Between Groups): This method calculates the distance between clusters as the average distance of all pairwise distances between points in each cluster. It balances within-group cohesion and between-group separation, providing a midpoint between other approaches (like single-linkage or complete-linkage).

   – Implication: This method is well-suited for finding clusters with moderate within-group similarity and maintaining separation between clusters. It avoids extreme closeness or excessive distance, making it ideal for datasets with moderately distinct groups.

 

 Agglomeration Schedule Interpretation

 

The Agglomeration Schedule provides a sequential look at how clusters merge at each stage, showing how cases combine from individual clusters into larger groupings based on increasing dissimilarity. Here’s an interpretation by stages:

 

1. Initial Stages (Early Combinations):

   – At the start, individual data points or very similar cases merge. For example, at Stage 1, clusters 1 and 3 merge with a very low coefficient of 8.000. This low coefficient indicates high similarity, meaning these clusters are nearly identical or very close in characteristics.

   – Implication: The initial merges represent cases that naturally group together because they share close characteristics, forming the foundation for subsequent, larger clusters. At these stages, the clustering is highly accurate because each merge is based on clear similarities.

 

2. Middle Stages (Moderate Combinations):

   – Around Stage 25 (coefficient 77.000), clusters are still merging based on moderate similarity, but there’s a notable increase in coefficients. This increase shows that clusters are becoming less homogeneous as broader groups start forming.

   – For example, Stages 10-25 involve coefficients between 32.000 and 77.000, indicating clusters that still retain identifiable traits but are less precise than in the initial stages.

   – Implication: The clusters formed in these stages are broader and incorporate cases with moderate differences. This phase captures more generalized clusters, which can represent key patterns or broader themes in the data.

 

3. Final Stages (Late Combinations):

   – In the final stages, coefficients increase sharply, with values such as 216.598 and 242.981 in Stages 52 and 53. This indicates that the remaining clusters are quite different from each other, but they are forced to merge to complete the clustering hierarchy.

   – Implication: At these stages, clusters contain more varied data points, which means that the merges represent less natural groupings. In hierarchical clustering, it’s common for the last merges to represent heterogeneous groupings, as the process attempts to organize all cases into fewer, broader clusters.

 

4. Key Takeaways:

   – Early Stages (High Similarity): Clusters are formed with cases that are highly similar, creating precise, cohesive groupings.

   – Middle Stages (Moderate Distinction): As clusters merge, they become less homogeneous but still retain enough similarity to represent broader patterns.

   – Late Stages (High Heterogeneity): The clustering combines increasingly dissimilar cases, producing broad, less cohesive groups. This indicates that the most meaningful clusters were likely formed earlier.

 

 Overall Interpretation

 

The clustering results suggest that your dataset has distinct natural groupings, particularly in the early stages where merges are based on strong similarities (low coefficients). As the process continues, clusters expand to include less similar cases, which eventually leads to less cohesive groupings in the final stages. This progression is typical in hierarchical clustering, where the initial stages provide the most interpretable and meaningful clusters, while later stages force merges that are less representative of actual similarities.

 

In sum, the Case Processing Summary and Agglomeration Schedule indicate a dataset with clear, distinct groupings at the start that become broader and more general as clustering progresses. This interpretation provides insight into the optimal clustering depth, suggesting that the most informative clusters are likely in the early to middle stages. This helps in understanding the structure within the data, identifying key groupings, and potentially determining where to stop merging for optimal cluster interpretation.

 

 

 

Descriptives
		N	Mean	Std. Deviation	Std. Error	95% Confidence Interval for Mean		Minimum	Maximum
						Lower Bound	Upper Bound
efficiency.	1	5	6.00	3.606	1.612	1.52	10.48	1	10
	2	2	4.00	2.828	2.000	-21.41	29.41	2	6
	3	4	7.25	2.500	1.250	3.27	11.23	4	10
	4	8	4.63	3.998	1.413	1.28	7.97	1	10
	5	4	5.25	.500	.250	4.45	6.05	5	6
	6	6	5.50	2.074	.847	3.32	7.68	4	9
	7	7	4.86	3.132	1.184	1.96	7.75	2	9
	8	9	4.78	3.598	1.199	2.01	7.54	1	10
	9	5	4.60	2.608	1.166	1.36	7.84	1	8
	10	4	7.75	2.217	1.109	4.22	11.28	5	10
	Total	54	5.35	2.985	.406	4.54	6.17	1	10
comfort.	1	5	4.80	2.387	1.068	1.84	7.76	2	8
	2	2	4.00	4.243	3.000	-34.12	42.12	1	7
	3	4	4.25	1.500	.750	1.86	6.64	2	5
	4	8	5.25	3.536	1.250	2.29	8.21	1	10
	5	4	4.75	3.304	1.652	-.51	10.01	1	9
	6	6	4.67	3.386	1.382	1.11	8.22	1	9
	7	7	4.29	2.928	1.107	1.58	6.99	1	9
	8	9	6.44	2.833	.944	4.27	8.62	2	10
	9	5	7.00	2.236	1.000	4.22	9.78	4	10
	10	4	5.50	3.873	1.936	-.66	11.66	1	10
	Total	54	5.24	2.920	.397	4.44	6.04	1	10
storage.	1	5	4.60	2.408	1.077	1.61	7.59	2	8
	2	2	4.50	4.950	3.500	-39.97	48.97	1	8
	3	4	7.00	2.708	1.354	2.69	11.31	3	9
	4	8	6.00	3.207	1.134	3.32	8.68	1	9
	5	4	6.75	3.202	1.601	1.66	11.84	4	10
	6	6	5.67	3.141	1.282	2.37	8.96	2	10
	7	7	6.00	3.317	1.254	2.93	9.07	2	10
	8	9	5.11	3.408	1.136	2.49	7.73	1	10
	9	5	7.60	2.074	.927	5.03	10.17	5	10
	10	4	6.00	3.559	1.780	.34	11.66	3	10
	Total	54	5.91	2.999	.408	5.09	6.73	1	10
combi-braking system (CBS).	1	5	5.40	2.302	1.030	2.54	8.26	3	9
	2	2	4.50	3.536	2.500	-27.27	36.27	2	7
	3	4	7.25	1.708	.854	4.53	9.97	5	9
	4	8	6.38	3.889	1.375	3.12	9.63	1	10
	5	4	8.50	1.915	.957	5.45	11.55	6	10
	6	6	6.33	3.386	1.382	2.78	9.89	2	10
	7	7	6.43	2.225	.841	4.37	8.49	3	9
	8	9	4.67	2.500	.833	2.74	6.59	1	7
	9	5	6.00	.707	.316	5.12	6.88	5	7
	10	4	3.50	1.732	.866	.74	6.26	1	5
	Total	54	5.91	2.708	.368	5.17	6.65	1	10
suspension.	1	5	4.40	3.435	1.536	.13	8.67	1	9
	2	2	5.50	4.950	3.500	-38.97	49.97	2	9
	3	4	5.25	2.217	1.109	1.72	8.78	3	8
	4	8	7.13	2.167	.766	5.31	8.94	3	9
	5	4	4.25	1.893	.946	1.24	7.26	3	7
	6	6	5.83	2.317	.946	3.40	8.26	3	9
	7	7	6.29	2.563	.969	3.91	8.66	2	10
	8	9	7.11	2.977	.992	4.82	9.40	3	10
	9	5	6.60	2.510	1.122	3.48	9.72	4	9
	10	4	5.75	3.948	1.974	-.53	12.03	1	9
	Total	54	6.06	2.716	.370	5.31	6.80	1	10
Affordability.	1	5	6.00	3.536	1.581	1.61	10.39	2	10
	2	2	4.00	4.243	3.000	-34.12	42.12	1	7
	3	4	6.00	1.414	.707	3.75	8.25	5	8
	4	8	5.25	3.536	1.250	2.29	8.21	1	10
	5	4	6.00	1.155	.577	4.16	7.84	5	7
	6	6	6.33	2.658	1.085	3.54	9.12	1	8
	7	7	3.14	2.673	1.010	.67	5.61	1	8
	8	9	6.11	2.713	.904	4.03	8.20	1	10
	9	5	6.00	2.121	.949	3.37	8.63	4	9
	10	4	4.00	2.449	1.225	.10	7.90	1	7
	Total	54	5.35	2.755	.375	4.60	6.10	1	10
Illumination.	1	5	5.60	3.647	1.631	1.07	10.13	2	10
	2	2	5.50	4.950	3.500	-38.97	49.97	2	9
	3	4	7.25	.957	.479	5.73	8.77	6	8
	4	8	7.13	3.441	1.217	4.25	10.00	1	10
	5	4	7.25	1.500	.750	4.86	9.64	6	9
	6	6	7.50	2.074	.847	5.32	9.68	5	10
	7	7	5.00	3.055	1.155	2.17	7.83	1	10
	8	9	4.67	2.398	.799	2.82	6.51	1	8
	9	5	6.40	3.912	1.749	1.54	11.26	1	10
	10	4	4.25	2.217	1.109	.72	7.78	2	7
	Total	54	6.02	2.891	.393	5.23	6.81	1	10

 

 

 Overall Descriptive Statistics Interpretation

The descriptive statistics summarize the performance of various characteristics across 54 cases, including efficiency, comfort, storage, combi-braking system (CBS), suspension, affordability, and illumination. Each characteristic’s statistics give insights into its distribution and variability within different clusters.

 

 Characteristic-by-Characteristic Analysis

 

1. Efficiency

   – Mean: The overall mean efficiency rating is 5.35, suggesting a moderate level of efficiency across cases.

   – Variability: The standard deviation of 2.985 indicates moderate variability in efficiency, with scores ranging from 1 to 10.

   – Cluster-Specific Trends:

     – Cluster 10 has the highest mean (7.75), indicating cases with above-average efficiency.

     – Clusters 2 and 4 show lower mean efficiency, around 4, suggesting a focus on characteristics other than high efficiency.

 

2. Comfort

   – Mean: The mean comfort level is 5.24, close to the dataset average.

   – Variability: With a standard deviation of 2.92, comfort levels vary widely across cases, with ratings spanning from 1 to 10.

   – Cluster-Specific Trends:

     – Cluster 9 has the highest mean comfort score of 7, indicating some cases prioritize comfort.

     – Cluster 2 has low comfort levels, with a wide confidence interval, suggesting less consistency.

 

3. Storage

   – Mean: The average storage rating is 5.91, indicating a slightly above-average prioritization of storage.

   – Variability: A standard deviation of 2.999 suggests considerable diversity in storage needs across cases.

   – Cluster-Specific Trends:

     – Clusters 3 and 9 show higher storage preferences, with means around 7, possibly indicating use cases with high storage requirements.

     – Cluster 2 has a wider confidence interval, suggesting variability in storage needs within that cluster.

 

4. Combi-Braking System (CBS)

   – Mean: The mean CBS rating is 5.91, in line with the average of the dataset.

   – Variability: Standard deviation is 2.708, indicating moderate variability.

   – Cluster-Specific Trends:

     – Cluster 5 scores highest (8.50), suggesting clusters that prioritize braking system features.

     – Clusters with lower CBS scores, like Cluster 10, may place less emphasis on braking systems.

 

5. Suspension

   – Mean: The suspension rating is 6.06, slightly above average, showing moderate emphasis on suspension quality.

   – Variability: Standard deviation is 2.716, suggesting some variability across clusters.

   – Cluster-Specific Trends:

     – Cluster 8 scores higher on suspension (7.11), indicating some clusters have a preference for quality suspension.

     – Clusters 5 and 2 exhibit lower suspension ratings, potentially prioritizing other characteristics.

 

6. Affordability

   – Mean: With a mean of 5.35, affordability is moderately prioritized.

   – Variability: A standard deviation of 2.755 indicates broad variability, with ratings from 1 to 10.

   – Cluster-Specific Trends:

     – Clusters 1 and 6 have slightly higher means around 6, suggesting some cases are more cost-conscious.

     – Cluster 7, with a low mean of 3.14, might reflect cases where affordability is less of a priority.

 

7. Illumination

   – Mean: The mean illumination rating is 6.02, suggesting slightly above-average importance on illumination.

   – Variability: A standard deviation of 2.891 indicates wide variability in lighting preferences.

   – Cluster-Specific Trends:

     – Clusters 4, 5, and 6 have higher illumination ratings, with means around 7, likely prioritizing well-lit environments or displays.

     – Cluster 10, with a mean of 4.25, indicates that some cases are less focused on illumination.

 

 Summary

 

In general, comfort, storage, suspension, and illumination show above-average mean values across clusters, suggesting that these features are commonly prioritized in the dataset. However, clusters like Cluster 10 for efficiency and Cluster 9 for comfort show that specific groups have unique priorities. The high variability (standard deviations between ~2 and ~3 for most features) indicates broad preferences across cases, with some characteristics being particularly diverse, such as storage and illumination. This analysis suggests that while there are general trends, there is significant diversity in how different cases prioritize these characteristics.

 

 

 

 

 

ANOVA
		Sum of Squares	df	Mean Square	F	Sig.
efficiency.	Between Groups	55.077	9	6.120	.645	.752
	Within Groups	417.238	44	9.483
	Total	472.315	53
comfort.	Between Groups	46.086	9	5.121	.555	.826
	Within Groups	405.784	44	9.222
	Total	451.870	53
storage.	Between Groups	40.665	9	4.518	.456	.896
	Within Groups	435.872	44	9.906
	Total	476.537	53
combi-braking system (CBS).	Between Groups	81.164	9	9.018	1.291	.269
	Within Groups	307.373	44	6.986
	Total	388.537	53
suspension.	Between Groups	51.658	9	5.740	.745	.666
	Within Groups	339.176	44	7.709
	Total	390.833	53
Affordability.	Between Groups	63.735	9	7.082	.920	.517
	Within Groups	338.579	44	7.695
	Total	402.315	53
Illumination.	Between Groups	73.456	9	8.162	.972	.476
	Within Groups	369.525	44	8.398
	Total	442.981	53

 

 

 ANOVA Results Interpretation

 

The ANOVA analysis examines whether there are significant differences across clusters for various characteristics, such as efficiency, comfort, storage, and others. Each characteristic’s F-statistic and significance level (p-value) indicate if clusters differ meaningfully in that characteristic. A p-value below 0.05 typically signifies a statistically significant difference across clusters, while higher values indicate that differences are likely due to random variation.

 

1. Efficiency

   – F-Statistic: 0.645, Sig.: 0.752

   – Interpretation: The high p-value (0.752) indicates no significant differences in efficiency across clusters. This suggests that efficiency is relatively consistent across the groups, and differences observed are likely due to chance.

 

2. Comfort

   – F-Statistic: 0.555, Sig.: 0.826

   – Interpretation: With a p-value of 0.826, comfort also shows no significant variation across clusters. This means that comfort levels are similarly distributed across the different clusters, showing no clear differentiation.

 

3. Storage

   – F-Statistic: 0.456, Sig.: 0.896

   – Interpretation: The very high p-value (0.896) suggests no meaningful differences in storage preferences across clusters. This implies that storage needs are consistent across groups, without a clear pattern distinguishing one cluster from another.

 

4. Combi-Braking System (CBS)

   – F-Statistic: 1.291, Sig.: 0.269

   – Interpretation: While CBS has a slightly higher F-statistic, its p-value of 0.269 remains above the 0.05 threshold, indicating no significant differences across clusters. However, CBS shows more variation between groups compared to other characteristics, hinting that some clusters might place a slightly higher emphasis on CBS features.

 

5. Suspension

   – F-Statistic: 0.745, Sig.: 0.666

   – Interpretation: Suspension has a p-value of 0.666, signifying no significant variation across clusters. Suspension preferences are spread similarly across groups, with no particular cluster standing out in terms of emphasis on suspension.

 

6. Affordability

   – F-Statistic: 0.920, Sig.: 0.517

   – Interpretation: Affordability also shows no significant cluster differences, with a p-value of 0.517. This indicates that affordability is valued consistently across clusters, and the observed variance is likely due to random distribution.

 

7. Illumination

   – F-Statistic: 0.972, Sig.: 0.476

   – Interpretation: With a p-value of 0.476, illumination does not vary significantly between clusters. This means that clusters do not differ notably in their preference for illumination, and any differences are likely by chance.

 

 Summary of ANOVA Results

 

Overall, the ANOVA results reveal that there are no statistically significant differences between clusters for any of the characteristics examined. The p-values for all characteristics are well above the 0.05 threshold, indicating that the clusters do not exhibit meaningful variations in terms of efficiency, comfort, storage, CBS, suspension, affordability, or illumination. This suggests that these attributes are distributed similarly across clusters, implying that the clustering structure may not be driven by these characteristics or that additional factors may be influencing the clustering pattern.

 

In sum, the ANOVA analysis highlights that differences in these characteristics across clusters are minimal, suggesting a homogeneous distribution of these attributes across the dataset. Future analyses could explore other variables or clustering methods to identify potential drivers of cluster distinctions.

 

 

 

 

 

ANOVA Effect Sizesa,b
		Point Estimate	95% Confidence Interval
			Lower	Upper
efficiency.	Eta-squared	.117	.000	.137
	Epsilon-squared	-.064	-.205	-.039
	Omega-squared Fixed-effect	-.063	-.200	-.038
	Omega-squared Random-effect	-.007	-.019	-.004
comfort.	Eta-squared	.102	.000	.111
	Epsilon-squared	-.082	-.205	-.071
	Omega-squared Fixed-effect	-.080	-.200	-.070
	Omega-squared Random-effect	-.008	-.019	-.007
storage.	Eta-squared	.085	.000	.077
	Epsilon-squared	-.102	-.205	-.111
	Omega-squared Fixed-effect	-.100	-.200	-.109
	Omega-squared Random-effect	-.010	-.019	-.011
combi-braking system (CBS).	Eta-squared	.209	.000	.271
	Epsilon-squared	.047	-.205	.121
	Omega-squared Fixed-effect	.046	-.200	.119
	Omega-squared Random-effect	.005	-.019	.015
suspension.	Eta-squared	.132	.000	.163
	Epsilon-squared	-.045	-.205	-.008
	Omega-squared Fixed-effect	-.044	-.200	-.008
	Omega-squared Random-effect	-.005	-.019	-.001
Affordability.	Eta-squared	.158	.000	.203
	Epsilon-squared	-.014	-.205	.040
	Omega-squared Fixed-effect	-.013	-.200	.039
	Omega-squared Random-effect	-.001	-.019	.005
Illumination.	Eta-squared	.166	.000	.213
	Epsilon-squared	-.005	-.205	.053
	Omega-squared Fixed-effect	-.005	-.200	.052
	Omega-squared Random-effect	-.001	-.019	.006
a. Eta-squared and Epsilon-squared are estimated based on the fixed-effect model.
b. Negative but less biased estimates are retained, not rounded to zero.

 

 

 

 

Effect sizes offer a way to quantify how much of the variability in each characteristic is due to clustering. For each measure:

 

– Eta-squared (η²): Represents the proportion of variance in each characteristic attributed to clusters. Higher values indicate stronger clustering effects.

– Epsilon-squared and Omega-squared: Adjustments to Eta-squared that provide less biased estimates. Negative values suggest minimal variance due to clustering.

 

1. Efficiency

   – Eta-squared: 0.117, indicating a low proportion of variance attributed to clustering.

   – Epsilon-squared and Omega-squared: Both values are negative, ranging between -0.205 and -0.038, reinforcing the finding that clustering does not meaningfully account for variability in efficiency.

 

2. Comfort

   – Eta-squared: 0.102, which is also low, indicating limited cluster-based variance in comfort.

   – Epsilon-squared and Omega-squared: These negative values (as low as -0.205) suggest that clustering has negligible impact on comfort, confirming that comfort is distributed similarly across clusters.

 

3. Storage

   – Eta-squared: 0.085, a minimal effect size, showing low cluster-based variance.

   – Epsilon-squared and Omega-squared: These values are negative and relatively small, suggesting storage differences across clusters are not meaningful.

 

4. Combi-Braking System (CBS)

   – Eta-squared: 0.209, a moderate effect size, which suggests CBS may have a somewhat stronger association with clusters than other characteristics.

   – Epsilon-squared and Omega-squared: Values for CBS are positive, up to 0.121, hinting at some differentiation across clusters. This suggests that CBS may vary more meaningfully with clustering.

 

5. Suspension

   – Eta-squared: 0.132, indicating a small effect size for suspension.

   – Epsilon-squared and Omega-squared: Near-zero to slightly negative values show that the variance due to clustering in suspension is minimal, with little differentiation across clusters.

 

6. Affordability

   – Eta-squared: 0.158, a slightly higher effect size, suggesting affordability could have a minor association with clustering.

   – Epsilon-squared and Omega-squared: Values are close to zero or slightly positive, indicating a small impact of clustering on affordability. Clusters may slightly differ in affordability preferences.

 

7. Illumination

   – Eta-squared: 0.166, another modest effect size, implying a weak clustering effect.

   – Epsilon-squared and Omega-squared: Values near zero suggest that clustering may influence illumination ratings marginally but without strong separation between groups.

 

 Summary of Effect Size Findings

 

Overall, CBS, Affordability, and Illumination show slightly higher Eta-squared values, suggesting these characteristics might exhibit more clustering-related variance than others, albeit still at low to moderate levels. The negative or near-zero values in Epsilon-squared and Omega-squared for most characteristics confirm that clustering only weakly differentiates these variables. This analysis implies that clusters do not vary substantially in their characteristics and that further analysis or additional variables may be needed to identify meaningful subgroup differences.