feat(mcda) - mcda feature documentation added

src/features/mcda/readme.md

@@ -0,0 +1,78 @@
+## MCDA - Multi Criteria Decision Analysis
+
+MCDA is a basic tool to answer localization questions with GIS ("where better do or not to do something").
+Users have a need to perform a spatial analysis based on a variety of criteria.
+Previously Kontur allowed combining only 2 layers in bivariate-manager feature.
+In MCDA feature we can allow users to select as many layers as they need.
+
+## How to use
+
+To start using this feature user needs:
+
+- feature flag "bivariate_manager" is turned on for current user
+- flag "KONTUR_MCDA": true in localstorage
+
+### Activation
+
+In Toolbar user will see a MCDA button.
+By clicking it in activated prompt window user needs to enter JSON of next structure:
+
+```json
+{
+  "id": "MCDA",
+  "layers": [
+    {
+      "axis": ["population", "area_km2"],
+      "range": [0, 46200],
+      "sentiment": ["good", "bad"],
+      "coefficient": 1
+    },
+    {
+      "axis": ["hazardous_days_count", "one"],
+      "range": [0, 365],
+      "sentiment": ["good", "bad"],
+      "coefficient": 1
+    },
+    {
+      "axis": ["highway_length", "total_road_length"],
+      "range": [0, 1],
+      "sentiment": ["bad", "good"],
+      "coefficient": 1
+    }
+  ],
+  "colors": {
+    "good": "rgba(90, 200, 127, 0.5)",
+    "bad": "rgba(228, 26, 28, 0.5)"
+  }
+}
+```
+
+### JSON structure
+
+1. id - name of layer on layers panel and in layers related atoms (non required)
+2. colors - coloring of sentiments (required)
+3. layers - all configured layers in analysis (all fields are required)
+   - axis: [numerator: string, denominator: string] - numerator, denominator
+   - range: [min: number, max: number] - min, max values for specific axis
+   - sentiment: only 2 options here - ["bad", "good"] as deafult direction and ["good", "bad"] as a reversed
+   - coefficient: weight of layer in analysis, could be >=0 and <=1
+
+## How it calculates
+
+MCDA uses data from bivariate tiles that have large amount of indicators inside.
+In every hexagon for every layer from provided JSON it's calculted:
+
+```
+V_norm = (V_current - V_min)/(V_max - V_min)
+```
+
+Where `V_current = (numerator/denominator)`, V_min, V_max are values from the range field.
+If direction of layer is reversed (["good", "bad"]) `V_norm = 1 - V_norm`.
+And in the end:
+
+```
+V_total = Σ(V_norm(1) · weight(1), V_norm(2) · weight(2), V_norm(n) · weight(n)) / Σweight
+```
+
+V_total is a result value of MCDA analysis (it's >=0 and <= 1), it's interpolated in color from colors field of JSON, where 0 is bad color, 1 is good color.
+If value is out of [0,1] range, it's painted as transparent.