feat(challenges): [api] add implementation

msmendoza · Feb 27, 2025 · a77e1af · a77e1af
1 parent 036f95d
commit a77e1af
Show file tree

Hide file tree

Showing 4 changed files with 379 additions and 344 deletions.
diff --git a/challenge/api.py b/challenge/api.py
@@ -1,4 +1,9 @@
 import fastapi
+import joblib
+import pandas as pd
+from fastapi import Request
+from fastapi.responses import JSONResponse
+from sklearn.linear_model import LogisticRegression
 
 
 app = fastapi.FastAPI()
@@ -10,5 +15,10 @@ async def get_health() -> dict:
 
 
 @app.post("/predict", status_code=200)
-async def post_predict() -> dict:
-    return
+async def post_predict(req: Request) -> JSONResponse:
+    features = pd.DataFrame((await req.json())["flights"])
+
+    model: LogisticRegression = joblib.load("data/model.joblib")
+    model.predict(features)
+
+    return JSONResponse(content={"predictions": features.to_dict()})
diff --git a/challenge/exploration.ipynb b/challenge/exploration.ipynb
diff --git a/data/reg_model.pkl b/data/reg_model.pkl
diff --git a/docs/challenge.md b/docs/challenge.md
@@ -12,3 +12,5 @@ repo and this repo only updates associated Docker images.
 1. `training_data` var was not being used.
 2. Although some better cols where found after some improvements I left the ones fixed in the
 tests.
+3. I chosed the Logistic Regression model because it is a simple model in comparison to the
+XGBoost model and performances are pretty similar > low processing costs.