arehman10 · arehman10 · May 25, 2025
diff --git a/README.md b/README.md
@@ -0,0 +1,52 @@
+# ISIC Classification Utilities
+
+This repository contains a simple ISIC Rev.4 classifier and a small demo app
+built with Streamlit. The classifier maps free‑text firm activity descriptions
+to four‑digit ISIC codes using token overlap and fuzzy string matching.
+
+## Installation
+
+Create a virtual environment and install the dependencies:
+
+```bash
+python -m venv .venv
+source .venv/bin/activate
+pip install -r requirements.txt
+```
+
+Alternatively, you can install the minimal packages directly:
+
+```bash
+pip install pandas streamlit
+```
+
+## Running the Streamlit app
+
+Launch the web interface with:
+
+```bash
+streamlit run isic_app.py
+```
+
+The app lets you classify a single activity description or upload a CSV file
+(with a `d1a1x` column) and obtain the predicted ISIC codes. The processed file
+can be downloaded directly from the app.
+
+## Command line usage
+
+The `isic_classifier.py` module can also be run from the command line:
+
+```bash
+python isic_classifier.py input.csv output.csv
+```
+
+The input CSV must contain a `d1a1x` column. The script will produce a copy of
+the file with `isic_code` and `match_score` columns added.
+
+## Testing
+
+Run the tests with:
+
+```bash
+pytest -q
+```
diff --git a/isic.csv b/isic.csv
@@ -0,0 +1,20 @@
+0111;Growing of cereals (except rice), leguminous crops and oil seeds
+0112;Growing of rice
+0113;Growing of vegetables and melons, roots and tubers
+0114;Growing of sugar cane
+0116;Growing of fibre crops
+0311;Marine fishing
+0510;Mining of hard coal
+1010;Processing and preserving of meat
+1071;Manufacture of bakery products
+1811;Printing
+2011;Manufacture of basic chemicals
+4610;Wholesale on a fee or contract basis
+4711;Retail sale in non-specialized stores with food, beverages or tobacco predominating
+5510;Short term accommodation activities
+5610;Restaurants and mobile food service activities
+6201;Computer programming activities
+7020;Management consultancy activities
+8610;Hospital activities
+9311;Operation of sports facilities
+9602;Hairdressing and other beauty treatment
diff --git a/isic_app.py b/isic_app.py
@@ -0,0 +1,39 @@
+import io
+import pandas as pd
+import streamlit as st
+
+from isic_classifier import load_isic, classify
+
+st.title("ISIC Rev.4 Classifier")
+
+isic_data = load_isic()
+
+st.header("Classify a single activity")
+text = st.text_input("Activity description")
+if st.button("Classify"):
+    if text:
+        code, score = classify(text, isic_data)
+        st.write(f"**ISIC code:** {code}")
+        st.write(f"**Match score:** {score:.2f}")
+    else:
+        st.warning("Please enter a description")
+
+st.header("Classify activities from a CSV file")
+file = st.file_uploader("Upload CSV with a 'd1a1x' column", type=["csv"])
+if file is not None:
+    df = pd.read_csv(file)
+    if "d1a1x" not in df.columns:
+        st.error("Column 'd1a1x' not found in uploaded file")
+    else:
+        codes = []
+        scores = []
+        for desc in df["d1a1x"].astype(str):
+            code, score = classify(desc, isic_data)
+            codes.append(code)
+            scores.append(round(score, 2))
+        df["isic_code"] = codes
+        df["match_score"] = scores
+        st.dataframe(df)
+        csv = df.to_csv(index=False).encode("utf-8")
+        st.download_button("Download results", csv, "isic_results.csv", "text/csv")
+
diff --git a/isic_classifier.py b/isic_classifier.py
@@ -0,0 +1,65 @@
+import csv
+import re
+import sys
+from difflib import SequenceMatcher
+from typing import List, Tuple
+
+
+def load_isic(path: str = "isic.csv") -> List[Tuple[str, str]]:
+    """Load ISIC code descriptions from a semicolon separated file."""
+    with open(path, newline='', encoding='utf-8') as f:
+        reader = csv.reader(f, delimiter=';')
+        return [(code.strip(), desc.strip()) for code, desc in reader]
+
+
+def _tokenize(text: str) -> set:
+    """Return a set of word tokens from the given text."""
+    return set(re.findall(r"\b\w+\b", text.lower()))
+
+
+def _ratio(a: str, b: str) -> float:
+    """Return a fuzzy string matching ratio between 0 and 1."""
+    return SequenceMatcher(None, a.lower(), b.lower()).ratio()
+
+
+def classify(activity: str, isic_data: List[Tuple[str, str]]) -> Tuple[str, float]:
+    """Return the ISIC code with the best combined similarity score."""
+    tokens = _tokenize(activity)
+    best_code = None
+    best_score = 0.0
+    for code, desc in isic_data:
+        desc_tokens = _tokenize(desc)
+        if not desc_tokens:
+            continue
+        token_overlap = len(tokens & desc_tokens) / len(desc_tokens)
+        ratio = _ratio(activity, desc)
+        score = 0.5 * token_overlap + 0.5 * ratio
+        if score > best_score:
+            best_score = score
+            best_code = code
+    return best_code, best_score
+
+
+def classify_file(input_csv: str, output_csv: str, column: str = "d1a1x", isic_path: str = "isic.csv") -> None:
+    """Classify activities in a CSV file and write results to a new CSV."""
+    isic_data = load_isic(isic_path)
+    with open(input_csv, newline='', encoding='utf-8') as f:
+        reader = csv.DictReader(f)
+        rows = list(reader)
+
+    for row in rows:
+        code, ratio = classify(row[column], isic_data)
+        row['isic_code'] = code
+        row['match_score'] = f"{ratio:.2f}"
+
+    with open(output_csv, 'w', newline='', encoding='utf-8') as f:
+        writer = csv.DictWriter(f, fieldnames=list(rows[0].keys()))
+        writer.writeheader()
+        writer.writerows(rows)
+
+
+if __name__ == "__main__":
+    if len(sys.argv) == 3:
+        classify_file(sys.argv[1], sys.argv[2])
+    else:
+        print("Usage: python isic_classifier.py input.csv output.csv")
diff --git a/requirements.txt b/requirements.txt
@@ -0,0 +1,2 @@
+pandas
+streamlit
diff --git a/sample_activities.csv b/sample_activities.csv
@@ -0,0 +1,6 @@
+d1a1x
+Growing of rice
+Computer software development
+Restaurants
+Short term accommodation services
+Operation of sports stadium
diff --git a/test_classifier.py b/test_classifier.py
@@ -0,0 +1,26 @@
+import csv
+import os
+from isic_classifier import classify_file
+
+
+def test_classify_file(tmp_path):
+    input_csv = os.path.join(tmp_path, 'input.csv')
+    output_csv = os.path.join(tmp_path, 'output.csv')
+    # copy sample activities file
+    with open('sample_activities.csv', 'r', encoding='utf-8') as src:
+        data = src.read()
+    with open(input_csv, 'w', encoding='utf-8') as dst:
+        dst.write(data)
+
+    classify_file(input_csv, output_csv)
+
+    with open(output_csv, newline='', encoding='utf-8') as f:
+        reader = csv.DictReader(f)
+        results = list(reader)
+
+    codes = [row['isic_code'] for row in results]
+    assert codes[0] == '0112'
+    assert codes[1] == '6201'
+    assert codes[2] == '5610'
+    assert codes[3] == '5510'
+    assert codes[4] == '9311'