Creating Pickle For Lung Cancer Model

flask_backend/models/LungCancer/LungCancer.py

@@ -2,6 +2,7 @@
 import pandas as pd
 import numpy as np
 import matplotlib.pyplot as plt
+import joblib
 #Importing necessary modules from sklearn
 from sklearn.metrics import confusion_matrix
 from sklearn.metrics import classification_report
@@ -11,8 +12,7 @@
 
 #Importing Dataset GSE74777_series_matrix.txt.gz
 # Tab-separated file
-train_table = pd.read_csv(r"C:\Users\cpkbh\OneDrive\Desktop\PredictMod\GSE74777_series_matrix.csv"
-)  
+train_table = pd.read_csv("GSE74777_series_matrix.csv")
 # ignores lines starting with !
 print("Table Loaded")
 print(train_table.head())
@@ -32,6 +32,13 @@
 X = train_table.drop(["Responder"], axis=1)
 y = train_table["Responder"]
 
+print(train_table.head(2))
+train_table.head(1).to_csv("train_table_head.csv", index=False)
+print('Printing json....')
+train_table.head(2).to_json("testInput.json", orient='records')
+
+train_table =train_table.iloc[2:].reset_index(drop=True)  # drop first two rows (metadata)
+
 # ---- Scale features ----
 scaler = StandardScaler()
 X_scaled = scaler.fit_transform(X)
@@ -48,19 +55,21 @@
 )
 
 print("Ready to Go")
-
+print(X_test[0,:])
 #Viewing performance metrics of Random Forest Classifier
-
-RF = RandomForestClassifier(random_state=123)
-RF.fit(X_train,y_train)
-print(f'RandomForestClassifier train score: {RF.score(X_train,y_train)}')
-print(f'RandomForestClassifier test score:  {RF.score(X_test,y_test)}')
-print(confusion_matrix(y_test, RF.predict(X_test)))
-print(classification_report(y_test, RF.predict(X_test)))
+rf = RandomForestClassifier(random_state=123)
+rf.fit(X_train,y_train)
+print(f'RandomForestClassifier train score: {rf.score(X_train,y_train)}')
+print(f'RandomForestClassifier test score:  {rf.score(X_test,y_test)}')
+print(confusion_matrix(y_test, rf.predict(X_test)))
+print(classification_report(y_test, rf.predict(X_test)))
 from sklearn.metrics import ConfusionMatrixDisplay
 
 # Get predictions ONCE (cleaner)
-y_pred = RF.predict(X_test)
+y_pred = rf.predict(X_test)
+
+data = {'classifier': rf, 'features': list(X.columns)}
+joblib.dump(data, 'LungCancer_classifier_and_features.pickle') 
 
 # Plot confusion matrix
 ConfusionMatrixDisplay.from_predictions(
@@ -78,7 +87,7 @@
 from sklearn.metrics import roc_curve, auc
 
 # Get probability predictions (NOT just 0/1 labels)
-y_probs = RF.predict_proba(X_test)[:, 1]
+y_probs = rf.predict_proba(X_test)[:, 1]
 
 # Compute ROC curve
 fpr, tpr, thresholds = roc_curve(y_test, y_probs)

flask_backend/models/LungCancer/dockerfile

@@ -0,0 +1,14 @@
+FROM python:3.11-slim
+
+WORKDIR /app
+
+COPY requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Copy your dictionary pickle file and script
+COPY LungCancer_classifier_and_features.pickle .
+COPY main.py .
+
+EXPOSE 8080
+
+CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8080"]

flask_backend/models/LungCancer/main.py

@@ -0,0 +1,58 @@
+from fastapi import FastAPI, HTTPException
+from pydantic import BaseModel
+import joblib
+import pandas as pd
+from typing import Dict, Any
+
+app = FastAPI(title = "Lung Cancer Prediction API")
+
+#load model file
+data = joblib.load('LungCancer_classifier_and_features.pickle')
+
+# Unpack classifier and required feature column names
+model = data['classifier']
+expected_features = data['features']
+
+# Define flexible request payload (Dictionary matching feature names -> values)
+class PredictRequest(BaseModel):
+    inputs: Dict[str, Any]
+
+@app.get("/")
+def home():
+    return {
+        "status": "online", 
+        "expected_features": expected_features
+    }
+
+@app.post("/predict")
+def predict(request: PredictRequest):
+    try:
+        # Convert incoming JSON dict to a pandas DataFrame with exact column ordering
+        input_df = pd.DataFrame([request.inputs])
+        print("Received input: {input_df.to_dict(orient='records')[0]}")  # Log the received input
+        # Ensure all required features are present in the incoming payload
+        missing_cols = set(expected_features) - set(input_df.columns)
+        if missing_cols:
+            raise HTTPException(
+                status_code=400, 
+                detail=f"Missing feature columns: {list(missing_cols)}"
+            )
+
+        # Reorder columns to strictly match X.columns from training
+        input_df = input_df[expected_features]
+
+        # Predict using the extracted Random Forest model
+        prediction = model.predict(input_df)
+
+        # Optional: Get prediction probabilities if classification
+        probability = None
+        if hasattr(model, "predict_proba"):
+            probability = model.predict_proba(input_df).tolist()
+
+        return {
+            "prediction": int(prediction[0]) if hasattr(prediction[0], 'item') else prediction[0],
+            "probabilities": probability
+        }
+
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=str(e))

flask_backend/models/LungCancer/readme.txt

@@ -0,0 +1,14 @@
+python -m venv .venv
+
+
+source .venv/bin/activate
+Create requirements.txt with list of libraries
+pip install -r requirements.txt
+save the model into a pickle file using joblib
+
+created API server as main.python
+load model and create HTTP endpoint
+create dockerfile
+uvicorn main:app --reload --host 127.0.0.1 --port 8080
+
+curl -X POST "http://localhost:8080/predict" -H "Content-Type: application/json" -d @testInput.json

flask_backend/models/LungCancer/requirements.txt

@@ -0,0 +1,7 @@
+pandas
+numpy
+matplotlib
+scikit-learn
+fastapi
+pydantic
+uvicorn