Allye Survey Analysis
Allye Survey Analysis extends Allye Base with widgets and templates designed for practical, business-focused survey analytics—going beyond simple complaint counts.
If you are new to this style of analysis, see the end-to-end tutorial: Survey Analysis.
This package is built around three common questions:
- How are dissatisfaction items related?
Complaints often co-occur because they share a hidden root cause. You want to understand correlation structure and extract latent factors. - Which issues matter most for outcomes like overall satisfaction, churn, or resignation?
Not every complaint has the same business impact—prioritization requires driver/sensitivity analysis. - Which pain points are shared vs. segment-specific?
Different customer segments may have distinct frustrations. You want a clear, visual way to compare them.
To support these workflows, this package adds the widgets below on top of Allye Base.
Key Features
1. Quick Aggregation & Connected Visualization
The basic workflow (filtering rows/columns, aggregation, and interactive charts) is supported in Allye Base. See Allye Base for details.
2. Identify Latent Factors (Root Causes)
| Widget | Description |
|---|---|
| Factor Analysis | Extracts latent factors from many numeric survey items (EFA) and helps you interpret them via a scree plot, suitability metrics (KMO / Bartlett), loadings barplot, and biplot. Supports factor selection (Kaiser / fixed / variance explained), rotation (Varimax / Promax), and extraction (PAF / ML). Outputs factor scores appended to the data plus factor loadings and scoring coefficients tables (requires numeric, non-missing features). |
3. Quantify Driver Impact (Sensitivity Analysis)
Sensitivity / driver analysis is supported in Allye Base via:
| Widget | Description |
|---|---|
| Regression Analysis | Models a continuous outcome (e.g., NPS, satisfaction score) and produces an interpretable report with coefficients, confidence intervals (when applicable), and diagnostics. |
| Binary Analysis | Models a binary outcome (e.g., churn, resigned vs. retained) with evaluation metrics and interpretable effect sizes (odds ratios). |
Tip: Use factor scores from Factor Analysis as compact, less-collinear inputs when modeling overall outcomes.
4. Map Common vs. Unique Issues Across Segments
| Widget | Description |
|---|---|
| Correspondence Analysis | Performs correspondence analysis on two categorical variables (e.g., Segment × Issue) and visualizes their association as a low-dimensional biplot with a scree plot and contribution-to-inertia views. Supports multiple normalization modes and outputs a coordinates table (components + Variable/Value metadata) for downstream visualization and reporting. |
5. Survey-Focused Data Transformation & Visualization
| Widget | Description |
|---|---|
| Melt | Converts wide survey tables into a long format (id, item, value). Useful for quickly comparing many question items with the same chart/table logic. Supports choosing a unique row identifier, ignoring non-numeric features, excluding zero values, and customizing generated column names. |
| Transpose | Transposes a data table (swap rows/columns) to switch the analysis perspective (e.g., treat questions as rows). Supports generating new feature names generically or from a chosen variable, with an option to remove redundant instances. |
| Linear Projection | Interactive 2D projection for multivariate numeric data with connected selection. Supports Circular placement, PCA placement, and (when a suitable discrete target is available) LDA placement. Includes feature selection and “Suggest Features” (VizRank) to help find informative axes for exploration. |
6. Clustering & Segmentation
| Widget | Description |
|---|---|
| k-Means | Segments users/items into clusters for deeper analysis. Supports selecting a fixed number of clusters or searching a range and comparing silhouette scores; outputs annotated data with cluster labels and cluster centroids. |
| Hierarchical Clustering | Creates a dendrogram-based segmentation from a distance matrix and lets you cut/select clusters interactively; outputs selected/annotated data for follow-up analysis. |
Templates
To see a practical end-to-end workflow, refer to the Survey Analysis tutorial.