Upload README-GoogleAPI-dataset.md

README-GoogleAPI-dataset.md

@@ -0,0 +1,119 @@
+---
+pretty_name: EV Charging Stations in Finland (Google Places API)
+size_categories:
+- 1K<n<10K
+tags:
+- electric-vehicles
+- ev-charging
+- finland
+- geospatial
+- google-places
+---
+
+# EV Charging Stations in Finland (Google Places API)
+
+This dataset lists every public EV charging station in Finland that Google Places returns. Each row is one Google place of type `electric_vehicle_charging_station`. For each place it gives the location and address, opening hours and Google rating. It also lists the connectors at the site, with each connector's type, maximum charging power and, for most sites, live availability.
+
+
+| | |
+|---|---|
+| Rows | **4,958** charging locations |
+| Columns | 17 |
+| Geography | Finland, including Åland: 554 towns, 1,092 postal codes |
+| Source | Google Places API (New) |
+| Snapshot | Availability data is timestamped **17 Sep 2026, 06:00–07:10 UTC** (09:00–10:10 Helsinki time) |
+| Primary key | `id`, the Google Place ID. Unique, with no duplicate rows. |
+| Coordinates | WGS84. `lat` 59.82–70.08, `lng` 19.86–30.94. All points are inside Finland. |
+
+## Files
+
+| File | Size | Contents |
+|---|---|---|
+| `ev_finland.csv` | 2.7 MB | Flattened table, one row per place. **This README describes this file.** |
+| `ev_finland_google_api.json` | 6.4 MB | The raw Places API records with nested fields (`displayName`, `location`, `regularOpeningHours`, `evChargeOptions`, `postalAddress`, …). This is the file the Hugging Face viewer shows. |
+
+In the CSV, `name` comes from `displayName.text`, and `lat`/`lng` come from `location`. `openingHours` is `regularOpeningHours.weekdayDescriptions` joined with `; `. `connectorCount` and `connectors` come from `evChargeOptions`. `postalCode` and `regionCode` come from `postalAddress`.
+
+## Quick start
+
+```python
+import pandas as pd, json
+
+df = pd.read_csv("hf://datasets/data-sci-project/ev-count-google-api/ev_finland.csv",
+                 dtype={"postalCode": str})        # keep leading zeros, e.g. "00580"
+
+# One row per connector group
+conn = (df.dropna(subset=["connectors"])
+          .assign(connectors=lambda d: d.connectors.map(json.loads))
+          .explode("connectors"))
+conn = pd.concat([conn[["id", "name"]].reset_index(drop=True),
+                  pd.json_normalize(conn.connectors.tolist())], axis=1)
+
+# Highest power per site. A site counts as a fast charger if it reaches at least 50 kW.
+df["max_kw"] = df.id.map(conn.groupby("id").maxChargeRateKw.max())
+fast = df[df.max_kw >= 50]                          # 1,337 sites
+```
+
+---
+
+## Columns
+
+| Column | Type | Filled | Distinct | Description |
+|---|---|---|---|---|
+| `id` | str | 4,958 | 4,958 | Google Place ID (`ChIJ…`, 27 chars). |
+| `name` | str | 4,958 | 192 | Display name. This is almost always a **brand label** such as "Plugit Charging Station", not the name of the site. |
+| `primaryType` | str | 4,958 | 1 | Always `electric_vehicle_charging_station`. |
+| `primaryTypeDisplayName` | str | 4,958 | 1 | Always `Electric vehicle charging station`. |
+| `pureServiceAreaBusiness` | bool | 25 | 1 | `False` on 25 rows and empty elsewhere. Carries no information. |
+| `formattedAddress` | str | 4,958 | 4,670 | Full address, e.g. `Kauppamiehentie 1, 21710 Pargas`. |
+| `postalCode` | str | 4,909 | 1,092 | Finnish postal code. Empty on 49 rows. |
+| `regionCode` | str | 4,958 | 1 | Always `FI`. |
+| `lat`, `lng` | float | 4,958 | 4,953 | Coordinates. |
+| `rating` | float | 1,508 | 37 | Average Google rating, 1.0–5.0. |
+| `userRatingCount` | float | 1,508 | 45 | Number of ratings, 1–2,880. |
+| `priceLevel` | float | **0** | — | Always empty. |
+| `priceRange` | float | **0** | — | Always empty. |
+| `connectorCount` | float | 4,273 | 64 | Total connectors at the place, 1–367. It always equals the sum of `count` in `connectors`. |
+| `connectors` | JSON str | 4,273 | 1,466 | List of connector groups. See below. |
+| `openingHours` | str | 4,180 | 71 | Weekly hours as text, e.g. `Monday: Open 24 hours; Tuesday: …`. |
+
+### The `connectors` field
+
+This column is a JSON list with one object per connector type and power level at the site. For example:
+
+```json
+[{"type": "EV_CONNECTOR_TYPE_CCS_COMBO_2", "maxChargeRateKw": 250, "count": 4,
+  "availableCount": 4, "outOfServiceCount": 0,
+  "availabilityLastUpdateTime": "2026-09-17T07:05:00Z"},
+ {"type": "EV_CONNECTOR_TYPE_TYPE_2", "maxChargeRateKw": 22, "count": 2}]
+```
+
+| Key | Present in | Meaning |
+|---|---|---|
+| `type` | all 5,673 groups | Connector standard (see table below). |
+| `maxChargeRateKw` | 5,672 groups | Maximum power of this connector group, in kW. |
+| `count` | all | Number of connectors in the group. |
+| `availableCount` | 5,112 groups (90%) | Connectors free at the snapshot time. |
+| `outOfServiceCount` | 5,112 groups | Connectors reported broken at the snapshot time. |
+| `availabilityLastUpdateTime` | 5,112 groups | UTC time when the availability was last updated. |
+
+---
+
+## Known issues
+
+1. **Brand names are inconsistent.** Casing and suffixes vary, e.g. `Virta`, `VIRTA`, `Virta Global`, `Liikennevirta`, and `korttelilataus` alongside `Korttelilataus`. Normalise names before grouping by brand:
+   ```python
+   brand = (df.name.str.lower()
+              .str.replace(r" charging station| global|\.fi|-lataus| oy.*", "", regex=True)
+              .str.strip().replace({"liikennevirta": "virta"}))
+   ```
+2. **The same site sometimes appears twice.** 8 rows share 4 coordinate pairs. Helsinki Airport P2 appears as two places, each with 246 connectors. Remove duplicates before summing connectors by area.
+3. **Some postal codes and addresses are bad.** 49 rows have no `postalCode`. Two codes are invalid: `115610` (Lahti) and `NR9 3LL` (Lappeenranta). 217 addresses contain only a postal code and town. A few are garbled, e.g. `unknown, -, 20520 Turku`, `Hanko`, and an address containing "SWEDEN" for a site in Kuusisto. For location, use `lat`/`lng`.
+4. **Addresses use different formats.** Åland addresses put the house number first and end in `Åland` (`8, Nyhemsvägen 22150, Åland`). The regex `\d{5},?\s+([^,]+)` pulls the town from 4,955 of the 4,958 addresses.
+5. **Some power values have floating-point noise.** 88 connector groups have values like `13.800000190734863`. Round `maxChargeRateKw` to 1 decimal place.
+6. **Opening hours need cleaning before parsing.** 24 rows write all-day opening as `12:00 AM - 11:59 PM` instead of `Open 24 hours`. 127 rows use thin spaces (U+2009) and narrow no-break spaces (U+202F) around the times.
+
+
+## Note
+
+- This was merged with data from Digitraffic, and the merged data is included in another repo.