Citizen Segments | CSIS Dashboard

Locality Filter All Mindanao records

Showing 4500 of 4500 respondent records. Predictive model scores remain the full-sample reference unless retraining is explicitly run.

Region LGU / Municipality

What the Segments Mean

Respondents are grouped by similar profile signals such as demographic background, housing and living conditions, and related survey responses. Each segment is then compared with the service-stage pattern to show how citizen characteristics relate to awareness, availment, satisfaction, and need for action.

Segment Size

Segment	Respondents	%
Segment A	270	6.0
Segment B	1288	28.6
Segment C	1358	30.2
Segment D	874	19.4
Segment E	710	15.8

Segment A

Younger or single citizen profile

270 respondents · 6.0% of records

Relationship insight: Segment A has lower awareness than the service average, suggesting that communication reach may be weaker for this citizen group.

Segment B

Citizen profile group

1288 respondents · 28.6% of records

Relationship insight: Segment B is close to the service average across the main stages, so its profile traits are more useful for describing the group than for identifying a large service-stage gap.

Segment C

Citizen profile group

1358 respondents · 30.2% of records

Relationship insight: Segment C is most different on Availed, where its result is higher than the service average.

Segment D

Home-secure profile

874 respondents · 19.4% of records

Relationship insight: Segment D is most different on Aware, where its result is higher than the service average.

Segment E

Home-secure profile

710 respondents · 15.8% of records

Relationship insight: Segment E is close to the service average across the main stages, so its profile traits are more useful for describing the group than for identifying a large service-stage gap.

Segment A

Younger or single citizen profile

Cluster Characteristics

Place of Work (A8=95) - Not Applicable 96.67%
Civil Status (A4=1) - Single 92.59%
Age Group (A3.1=1) - 18→24 91.11%
Enrolled in School? (A6=1) - Yes 89.26%
Source of Electricty (B5=1) - ELECTRICITY OWN Connection 88.89%
Employment Status (A7=7) - Student (not working) 81.11%

Service Stage Profile

Aware 69.5%

Below average vs service average 73.8%

Availed 64.4%

Near average vs service average 63.0%

Satisfied 91.8%

Near average vs service average 93.0%

Action Needed 48.1%

Near average vs service average 46.8%

Segment B

Citizen profile group

Cluster Characteristics

Enrolled in School? (A6=2) - No 97.83%
Place of Work (A8=95) - Not Applicable 93.63%
Sex (A2=2) - Female 89.44%
Source of Electricty (B5=1) - ELECTRICITY OWN Connection 80.43%
Civil Status (A4=2) - Married 79.89%
4Ps Beneficiary (B1=2) - No 67.08%

Service Stage Profile

Aware 73.9%

Near average vs service average 73.8%

Availed 62.2%

Near average vs service average 63.0%

Satisfied 93.2%

Near average vs service average 93.0%

Action Needed 45.4%

Near average vs service average 46.8%

Segment C

Citizen profile group

Cluster Characteristics

Enrolled in School? (A6=2) - No 97.28%
Civil Status (A4=2) - Married 85.27%
Source of Electricty (B5=1) - ELECTRICITY OWN Connection 81.08%
Sex (A2=1) - Male 78.42%
Place of Work (A8=1) - W/in the barangay 75.48%
House Ownership (B2=1) - OWNER, OWNER-LIKE POSSESSION OF HOUSE AND LOT 68.85%

Service Stage Profile

Aware 75.5%

Near average vs service average 73.8%

Availed 66.2%

Above average vs service average 63.0%

Satisfied 94.0%

Near average vs service average 93.0%

Action Needed 47.9%

Near average vs service average 46.8%

Segment D

Home-secure profile

Cluster Characteristics

Enrolled in School? (A6=2) - No 95.88%
Source of Electricty (B5=1) - ELECTRICITY OWN Connection 91.99%
4Ps Beneficiary (B1=2) - No 89.24%
House Ownership (B2=1) - OWNER, OWNER-LIKE POSSESSION OF HOUSE AND LOT 77.69%
Sex (A2=1) - Male 64.53%
Employment Status (A7=1) - Working at least 40 hrs/wk 64.19%

Service Stage Profile

Aware 77.2%

Above average vs service average 73.8%

Availed 60.2%

Near average vs service average 63.0%

Satisfied 92.1%

Near average vs service average 93.0%

Action Needed 46.9%

Near average vs service average 46.8%

Segment E

Home-secure profile

Cluster Characteristics

Enrolled in School? (A6=2) - No 99.44%
Place of Work (A8=95) - Not Applicable 94.23%
Source of Electricty (B5=1) - ELECTRICITY OWN Connection 92.25%
House Ownership (B2=1) - OWNER, OWNER-LIKE POSSESSION OF HOUSE AND LOT 84.37%
4Ps Beneficiary (B1=2) - No 81.27%
Employment Status (A7=8) - Retired (not working) / Too old to work 70.14%

Service Stage Profile

Aware 72.7%

Near average vs service average 73.8%

Availed 62.2%

Near average vs service average 63.0%

Satisfied 93.7%

Near average vs service average 93.0%

Action Needed 45.6%

Near average vs service average 46.8%

Statistical Support

This section checks which profile variables meaningfully separate the citizen segments. Categorical variables use Chi-square, while age uses Kruskal-Wallis. Effect size shows the strength of the relationship, so larger values are more useful for explaining the segments.

Profile Variable	Method	Effect Size	p-value	Reader Meaning
Sex A2	Chi-square	0.543	<0.001	Strong relationship
Age A3	Kruskal-Wallis	0.512	<0.001	Strong relationship
Age Group A3.1	Chi-square	0.512	<0.001	Strong relationship
Relationship to HH Head A1	Chi-square	0.41	<0.001	Strong relationship
Civil Status A4	Chi-square	0.386	<0.001	Strong relationship
Highest Educational Attainment A5	Chi-square	0.324	<0.001	Strong relationship
4Ps Beneficiary B1	Chi-square	0.265	<0.001	Strong relationship
House Ownership B2	Chi-square	0.106	<0.001	Moderate relationship

Clustering Model Comparison

Candidate models test different profile-variable sets. The recommended row ranks highest by silhouette, with cluster balance used as supporting evidence. Service outcomes are still interpreted after clustering, not used to create the segments.

Rank	Model	Best k	Silhouette	Calinski-Harabasz	Davies-Bouldin	Balance	Features
1	Socio-housing core Recommended MCA + KMeans · 4500 respondents	5	0.358	2256.14	0.913	0.199	12 variables A1, A2, A3.1, A4, A5, A6, A7, A8, B1, B2, B5, B6
2	Housing-enriched MCA + KMeans · 4500 respondents	4	0.34	1561.264	1.116	0.173	10 variables A2, A3.1, A4, A5, B1, B2, B3, B4, B5, B6
3	Focused profile MCA + KMeans · 4500 respondents	3	0.333	1316.428	1.204	0.142	8 variables A2, A3.1, A4, A5, B1, B2, B5, B6
4	Full profile without IDs MCA + KMeans · 4500 respondents	4	0.324	1708.895	1.094	0.221	14 variables A1, A2, A3.1, A4, A5, A6, A7, A8, B1, B2, B3, B4, B5, B6

Segment Map

Model Selection

Profile Signal Map

How to Read the Advanced View

Segment Map: shows how respondents are positioned based on similar profile patterns. Points that appear closer together have more similar response characteristics.

Profile Signal Map: shows which demographic, housing, and background signals help explain the placement of the citizen segments.

Model Selection: supports the selected number of segments by checking where adding more groups gives less additional separation.

Best k

Silhouette

0.358

Calinski-Harabasz

2256.140

Davies-Bouldin

0.913

Thesis-Style Metric Interpretation

Interpretation. The clustering analysis identified five (5) optimal clusters within the dataset. The resulting Silhouette Score of 0.358 indicates moderate separation among clusters, suggesting that respondents within the same cluster share similar characteristics while still allowing some overlap with neighboring clusters. The Calinski-Harabasz Index of 2256.140 provides strong evidence of meaningful partitioning by comparing between-cluster variation with within-cluster variation. The Davies-Bouldin Index of 0.913 indicates acceptable cluster compactness and separation; lower values represent better cluster quality.

Academic Assessment

Silhouette = 0.358 (0.250-0.499) → Acceptable for social science and survey datasets where categorical responses often create overlapping groups.
Calinski-Harabasz = 2256.140 (> 2000) → Strong evidence of meaningful partitioning.
Davies-Bouldin = 0.913 (< 1) → Good indication of cluster quality.

Therefore, the five (5)-cluster solution can be considered statistically acceptable and suitable for interpretation, particularly in a CSIS survey context where perfect separation among citizen groups is rarely expected.