Saving & Loading Models with Pickle & Joblib

import pickle
from sklearn.linear_model import LogisticRegression
from sklearn.datasets import load_iris

X, y = load_iris(return_X_y=True)
model = LogisticRegression(max_iter=200)
model.fit(X, y)

# Save the fitted model to disk — note the "wb" (write, binary) mode
with open("iris_model.pkl", "wb") as f:
    pickle.dump(model, f)

print("Model saved to iris_model.pkl")

import pickle

# Load the model back — "rb" (read, binary) mode
with open("iris_model.pkl", "rb") as f:
    loaded_model = pickle.load(f)

# The loaded object behaves exactly like the original fitted model
prediction = loaded_model.predict(X[:1])
print(f"Prediction: {prediction}")

import joblib
from sklearn.ensemble import RandomForestClassifier
from sklearn.datasets import load_iris

X, y = load_iris(return_X_y=True)
model = RandomForestClassifier(n_estimators=100, random_state=42)
model.fit(X, y)

# joblib.dump() takes the object and a filename directly — no "with open()" needed
joblib.dump(model, "iris_model.joblib")
print("Model saved to iris_model.joblib")

import joblib

loaded_model = joblib.load("iris_model.joblib")
prediction = loaded_model.predict(X[:1])
print(f"Prediction: {prediction}")

import pandas as pd
import joblib
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.ensemble import RandomForestRegressor
from sklearn.pipeline import Pipeline

# 1. Load data
df = pd.read_csv("student_scores.csv")
features = ["StudyHours", "Attendance", "PrevScore", "AssignmentsDone"]
X = df[features]
y = df["FinalScore"]

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

# 2. Build ONE pipeline — scaler and model saved together as a single object
pipeline = Pipeline([
    ("scaler", StandardScaler()),
    ("model", RandomForestRegressor(n_estimators=200, random_state=42))
])
pipeline.fit(X_train, y_train)

print(f"Test R²: {pipeline.score(X_test, y_test):.3f}")

# 3. Save the WHOLE pipeline with joblib
joblib.dump(pipeline, "student_score_model.joblib")
print("Pipeline saved to student_score_model.joblib")

import joblib
import pandas as pd

# Load the saved pipeline — this is often a totally separate script/process
pipeline = joblib.load("student_score_model.joblib")

# A brand-new student, never seen during training
new_student = pd.DataFrame([{
    "StudyHours": 5.0,
    "Attendance": 88,
    "PrevScore": 70,
    "AssignmentsDone": 8
}])

predicted_score = pipeline.predict(new_student)[0]
print(f"Predicted final score: {predicted_score:.1f}")

import pandas as pd
import joblib
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.ensemble import RandomForestRegressor
from sklearn.pipeline import Pipeline

# 1. Load & split
df = pd.read_csv("student_scores.csv")
features = ["StudyHours", "Attendance", "PrevScore", "AssignmentsDone"]
X, y = df[features], df["FinalScore"]
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

# 2. Fit a pipeline
pipeline = Pipeline([("scaler", StandardScaler()), ("model", RandomForestRegressor(n_estimators=200, random_state=42))])
pipeline.fit(X_train, y_train)
print(f"Test R²: {pipeline.score(X_test, y_test):.3f}")

# 3. Save with joblib, versioned filename
MODEL_PATH = "student_score_model_v1.joblib"
joblib.dump(pipeline, MODEL_PATH)

# 4. Simulate a fresh process: load and predict
loaded_pipeline = joblib.load(MODEL_PATH)
new_student = pd.DataFrame([{"StudyHours": 5.0, "Attendance": 88, "PrevScore": 70, "AssignmentsDone": 8}])
print(f"Predicted score: {loaded_pipeline.predict(new_student)[0]:.1f}")