| --- |
| title: ImpactLedger |
| emoji: 🌱 |
| colorFrom: blue |
| colorTo: green |
| sdk: static |
| app_file: index.html |
| pinned: false |
| license: mit |
| short_description: Program impact and cost simulator. |
| --- |
| |
| # ImpactLedger |
|
|
| ### Program Impact Intelligence Studio | Hugdrigo Portfolio Project #106 |
|
|
| **ImpactLedger** is a self-contained, interactive educational simulator for comparing the costs, reach, observed outcome-rate changes and *assumed* attribution of fictional social programs. It includes a signed incremental-outcome model, interactive scenarios, reproducible Monte Carlo-style assumption shocks, a locally editable registry and portable JSON/CSV exports. It runs directly in the browser, with no backend or paid API. |
|
|
| **Short description:** Program impact and cost simulator. |
|
|
| > **Critical interpretation note:** These are synthetic, invented programs and illustrative calculations. A before-and-after rate difference does **not** establish causal impact. The attribution percentages are user-defined assumptions, not estimated causal effects. The simulated outcome units are only comparable because this artificial dataset treats them as standardized. Never aggregate real health, education and employment outcomes as if they were interchangeable without a validated common framework. This is not a randomized evaluation, a causal-inference system, a certified SROI calculation, or financial advice. |
|
|
| ## Publishing to Hugging Face |
|
|
| 1. Open [Create a new Space](https://huggingface.co/new-space). |
| 2. Set owner to **`Hugdrigo`** and name to **`impactledger-program-intelligence`**. |
| 3. Select **Static** SDK, **Blank** template, **Public** visibility, and **MIT** license. |
| 4. Extract the downloaded ZIP and upload the three files to the **Space repository root**, not a subfolder: `README.md`, `index.html` and `LICENSE`. |
| 5. Commit the changes and open the **App** tab. |
|
|
| The `README.md` front matter uses Hugging Face-compatible colors: `colorFrom: blue`, `colorTo: green`. The Space does not require Docker, GPUs, a billing account, package installs, network access or secrets. |
|
|
| ## Features |
|
|
| | Area | Included functions | |
| |---|---| |
| | Impact overview | Budget, reach, signed modeled incremental outcomes, cost per positive outcome, budget allocation and program comparison | |
| | Scenario studio | Four presets, four interactive controls, scenario vs reference outcomes, per-program cost efficiency | |
| | Program registry | Eight fictional example programs, add/edit (up to 120), search, archive/restore, input bounds, CSV export, JSON backup and restore | |
| | Uncertainty lab | 400 reproducible seeded draws, synthetic P10/P50/P90, outcome-sensitivity comparison and methodology warnings | |
| | Methodology | Exact equations, assumptions, interpretation cautions and program catalog | |
| | UI | Responsive layout, dark mode, local-first storage and offline operation | |
|
|
| ### Scenario presets |
|
|
| - **Baseline:** no adjustments. |
| - **Funding squeeze:** 15% reduced reach, -3 percentage points outcome rate, -20% budget, -5 percentage points attribution. |
| - **Delivery boost:** +35% reach, +6 percentage points outcome rate, +20% budget, +5 points attribution. |
| - **Measurement stress:** -10% reach, -8 points outcome rate, +15% cost, -20 points attribution. |
|
|
| Controls support reach -60% to +100%, measured outcome-rate changes -20 to +20 percentage points, budget -40% to +100%, and attribution -40 to +40 percentage points. Percentages are clamped to 0–100% where applicable. |
|
|
| ## Transparent equations |
|
|
| For one fictional program: |
|
|
| ```text |
| Adjusted reach = reach × (1 + reach_change_percent / 100) |
| Adjusted budget = budget × (1 + budget_change_percent / 100) |
| Adjusted follow-up rate = clamp(follow_up_rate + rate_change_pp, 0, 100) |
| Adjusted attribution = clamp(attribution_percent + attribution_change_pp, 0, 100) |
| Observed change (percentage points) = adjusted_follow_up_rate - baseline_rate |
| Modeled signed incremental outcomes = adjusted_reach × observed_change / 100 × adjusted_attribution / 100 |
| Cost per positive incremental outcome = adjusted_budget / modeled_incremental_outcomes, only when modeled_incremental_outcomes > 0 |
| ``` |
|
|
| Negative outcome-rate differences yield negative additional outcomes. Non-positive incremental outcomes show cost effectiveness as `N/A` rather than misleading positive efficiency. Portfolio totals assume that programs and outcome units do not overlap. |
|
|
| ### Synthetic uncertainty draws |
|
|
| The same deterministic seeded generator yields 400 draws with bounded invented reach, follow-up rate and attribution shocks. Percentiles are descriptive summaries of the *chosen simulation*, **not** calibrated uncertainty estimates, confidence intervals, counterfactual results or claims of actual impact. |
|
|
| ### Local data and safety |
|
|
| - All computation runs in the browser; no third-party scripts, external tracking, cloud storage or network APIs. |
| - Data saves in browser localStorage when supported, and can be backed up and restored using JSON. |
| - JSON import checks backup schema/version, unique numeric IDs, field types, ranges, a 2 MB file-size limit and up to 120 records. |
| - Imported program names are HTML-escaped before display. CSV exports quote cells and protect against spreadsheet formula injection while preserving legitimate negative numbers. |
| - Browser storage is not encrypted. Do not enter sensitive participant or beneficiary information. |
| - Archiving preserves program records in saved state; `Reset demo` is a deliberate, confirmed replacement of local data. Back up first. |
|
|
| ## Suggested evaluation workflow |
|
|
| Start with fictional examples, replace each assumption only with authorized aggregated data, inspect program-by-program changes, document your counterfactual/evaluation design independently, and use the scenario studio to test how strongly results depend on assumptions. Real impact reports require verified outcomes, evidence grading and causal identification. |
|
|
| ## Technical implementation |
|
|
| - Standalone `index.html` with inline accessible HTML, CSS, JavaScript and SVG; no dependencies or build step. |
| - MIT license in `LICENSE`. |
| - Designed for Hugging Face Static Spaces. |
| - All findings and comparison values are illustrative; the application is a portfolio demonstration, not an outcome assessment service. |
|
|
| © 2026 Hugdrigo. Project #106. |
|
|
