Your first machine learning model.

Today you'll train a real machine learning model that predicts flat prices, in about 15 lines of Python. You'll see the full process that professionals follow: split the data, train, predict and measure.

The ML recipe

1. Collect data (features + label).
2. Split into training data (to learn) and test data (to check).
3. Choose a model and train it.
4. Predict on the test data.
5. Measure how good the predictions are.

Why split? If you test a student with the exact questions they practised, a high score proves nothing. Test data is the "new exam paper".

Model 1: predicting a number (regression)

import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_absolute_error

df = pd.DataFrame({
    "area_sqft": [900,1000,1100,1200,1300,1400,1500,1650,1800,2000,1250,950],
    "bedrooms":  [2,2,2,2,3,3,3,3,3,4,2,2],
    "price_lakh":[55,60,66,72,80,90,96,118,125,140,76,57],
})
X = df[["area_sqft", "bedrooms"]]   # features
y = df["price_lakh"]                # label

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.25, random_state=1)

model = LinearRegression()
model.fit(X_train, y_train)              # learn

pred = model.predict(X_test)             # predict
print("Average error (lakh):", mean_absolute_error(y_test, pred))
print("1600 sqft, 3 BHK:", model.predict(pd.DataFrame({"area_sqft":[1600],"bedrooms":[3]})))

Mean absolute error tells you, on average, how far predictions are from real prices. Smaller is better.

Model 2: predicting a category (classification)

Will a student pass? (1 = pass, 0 = fail)

from sklearn.tree import DecisionTreeClassifier
from sklearn.metrics import accuracy_score

df = pd.DataFrame({
    "hours_studied": [1,2,3,4,5,6,7,8,2,9,3,7],
    "attendance":    [50,60,65,70,80,85,90,95,55,98,75,88],
    "passed":        [0,0,0,1,1,1,1,1,0,1,0,1],
})
X = df[["hours_studied", "attendance"]]
y = df["passed"]
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.25, random_state=1)

clf = DecisionTreeClassifier(max_depth=2)
clf.fit(X_train, y_train)
print("Accuracy:", accuracy_score(y_test, clf.predict(X_test)))

Tiny data warning

We used tiny datasets so the code is easy to read. Real models need much more data, and their results here will change a lot with small changes. That's normal, and an important lesson in itself.