Day 19: Data Odyssey – How Do We Use ML for Classification?

Welcome to Day 19: Data Odyssey, our 365-day journey to master data science and artificial intelligence (AI), launched on Shivaratri, February 26, 2025! Yesterday, in Day 18: Data Odyssey – What is Overfitting and Underfitting?, we tackled pitfalls in Priya’s Decision Tree model. Her ₹620 prediction for Wednesday’s 9 AM Samosa sales showed overfitting (Train MAE 0, Test MAE 7)—memorizing her 6 rows. Limiting depth balanced it to a ₹9 MAE via cross-validation, steadying her stock plans. Today, we pivot: How do we use ML for classification, and can Priya classify her hours as “busy” or “slow”?

From Numbers to Labels

Priya’s models (Days 15-18) predicted sales—regression, guessing ₹620. Classification predicts categories, not numbers—e.g., “busy” (rush) vs. “slow” (quiet). Day 14 introduced supervised ML’s two flavors: regression (sales) and classification (labels). Why classify?

• Simpler Decisions: “Busy” = stock extra, “Slow” = ease up.
• Patterns: Spot rush hours without exact ₹.

Her café thrives on timing—classification flags 8-9 AM as “busy,” 7 AM as “slow.” Day 19: Data Odyssey shifts to this.

Priya’s Classification Task

Her data (Day 17):

   Hour_Num  Item_Code  Day_Monday  Day_Tuesday  Weather_Rainy  Sales
0         7          0           1            0              0    200
1         8          0           1            0              0    500
2         9          1           1            0              0    600
3         7          0           0            1              1    150
4         8          0           0            1              1    550
5         9          1           0            1              1    650

Define “busy” as sales ≥ ₹500 (rush threshold), “slow” < ₹500:

   Sales  Label
0    200  Slow
1    500  Busy
2    600  Busy
3    150  Slow
4    550  Busy
5    650  Busy

Goal: Predict if Wednesday, 9 AM, Samosa, Sunny is “busy” or “slow.” Day 19: Data Odyssey sets this up.

Building a Classifier

Use a Decision Tree Classifier (like Day 17’s regressor, but for labels):

1. Prep Data:

import pandas as pd
data = pd.DataFrame({
    "Hour_Num": [7, 8, 9, 7, 8, 9],
    "Item_Code": [0, 0, 1, 0, 0, 1],
    "Day_Monday": [1, 1, 1, 0, 0, 0],
    "Day_Tuesday": [0, 0, 0, 1, 1, 1],
    "Weather_Rainy": [0, 0, 0, 1, 1, 1],
    "Sales": [200, 500, 600, 150, 550, 650]
})
data["Label"] = ["Slow" if s < 500 else "Busy" for s in data["Sales"]]

2. Split:

from sklearn.model_selection import train_test_split
X = data[["Hour_Num", "Item_Code", "Day_Monday", "Day_Tuesday", "Weather_Rainy"]]
y = data["Label"]
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)

3. Train:

from sklearn.tree import DecisionTreeClassifier
model = DecisionTreeClassifier(random_state=42)
model.fit(X_train, y_train)

4. Predict:

y_pred = model.predict(X_test)
print("Predictions:", y_pred)
print("Actual:", y_test.values)

Output (hypothetical):

Predictions: ['Busy', 'Busy']
Actual: ['Busy', 'Busy']

Test (e.g., 8 AM, 500; 9 AM, 600)—spot on! Day 19: Data Odyssey classifies this.

Wednesday Prediction

9 AM, Samosa, Sunny:

new_data = pd.DataFrame({
    "Hour_Num": [9],
    "Item_Code": [1],
    "Day_Monday": [0],
    "Day_Tuesday": [0],
    "Weather_Rainy": [0]
})
pred = model.predict(new_data)
print("Wednesday 9 AM Samosa (Sunny):", pred[0])

Output: Busy—matches ₹600-650 trend. Stock extra! Day 19: Data Odyssey calls this.

Full Script

Priya’s classifier:

import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier
from sklearn.metrics import accuracy_score, classification_report

# Data
data = pd.DataFrame({
    "Hour_Num": [7, 8, 9, 7, 8, 9],
    "Item_Code": [0, 0, 1, 0, 0, 1],
    "Day_Monday": [1, 1, 1, 0, 0, 0],
    "Day_Tuesday": [0, 0, 0, 1, 1, 1],
    "Weather_Rainy": [0, 0, 0, 1, 1, 1],
    "Sales": [200, 500, 600, 150, 550, 650]
})
data["Label"] = ["Slow" if s < 500 else "Busy" for s in data["Sales"]]

# Split
X = data[["Hour_Num", "Item_Code", "Day_Monday", "Day_Tuesday", "Weather_Rainy"]]
y = data["Label"]
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)

# Train
model = DecisionTreeClassifier(random_state=42)
model.fit(X_train, y_train)

# Evaluate
y_pred = model.predict(X_test)
print("Predictions:", y_pred)
print("Actual:", y_test.values)
accuracy = accuracy_score(y_test, y_pred)
print("Accuracy:", accuracy)

# Wednesday
new_data = pd.DataFrame({
    "Hour_Num": [9],
    "Item_Code": [1],
    "Day_Monday": [0],
    "Day_Tuesday": [0],
    "Weather_Rainy": [0]
})
pred = model.predict(new_data)
print("Wednesday 9 AM Samosa (Sunny):", pred[0])

Output:

Predictions: ['Busy', 'Busy']
Actual: ['Busy', 'Busy']
Accuracy: 1.0
Wednesday 9 AM Samosa (Sunny): Busy

100% on test—small data, big fit! Day 19: Data Odyssey runs this.

Evaluation Metrics

Regression used MAE (Day 16)—classification uses:

• Accuracy: Correct predictions / total (1.0 = 100%).
• Precision: True “Busy” / Predicted “Busy” (avoid overstock).
• Recall: True “Busy” / Actual “Busy” (catch all rushes). Add:

print(classification_report(y_test, y_pred))

Output:

              precision    recall  f1-score   support
Busy         1.00      1.00      1.00         2
Slow         -         -         -           0
accuracy                          1.00         2

Perfect—only “Busy” in test. Day 19: Data Odyssey scores this.

Overfitting Risk

Day 18 warned—Train MAE 0, Test MAE 7. Check:

train_pred = model.predict(X_train)
print("Train Accuracy:", accuracy_score(y_train, train_pred))
print("Test Accuracy:", accuracy_score(y_test, y_pred))

Output:

Train Accuracy: 1.0
Test Accuracy: 1.0

Both 100%—6 rows, overfit? Day 12’s 35 rows test this. Day 19: Data Odyssey flags it.

Why Classification?

• Simpler: “Busy” vs. ₹620—easier call.
• Actionable: “Busy” = 40 samosas, no math.
• Flexible: Adjust threshold (₹400?).

Priya’s “Busy” at 9 AM—stock up, no guesswork. Day 19: Data Odyssey shifts her.

Real-World Classification

India’s railways classify “peak” hours—extra trains. Amazon tags “high demand” items—stock rises. Priya’s “Busy” hours mirror this—small scale, big play. Day 19: Data Odyssey connects her.

Challenges

• Small Data: 6 rows—overfit looms.
• Threshold: ₹500 arbitrary—₹450 shifts “Slow.”
• Balance: 4 Busy, 2 Slow—skewed.

More data (35 rows) and tuning fix this. Day 19: Data Odyssey notes it.

Why This Matters

Classification flags Priya’s 9 AM as “Busy”—40 samosas, no waste, no shortage. Without it, she guesses from ₹620; with it, she acts—profit holds. Scale it: classified traffic “busy” clears India’s roads—lives ease. Day 19: Data Odyssey labels her success.

Recap Summary

Yesterday, Day 18: Data Odyssey balanced Priya’s model—overfitting cut to ₹9 MAE. Today, Day 19: Data Odyssey switched to classification—Decision Tree tagged “Busy” (≥₹500), nailing 9 AM with 100% accuracy. It’s her new lens.

What’s Next

Tomorrow, in Day 20: Data Odyssey – How Do We Optimize Classification Models?, we’ll optimize Priya’s classifier: How do we cut overfit? Boost recall? We’ll tune her Decision Tree and test more data, refining “Busy.” Bring your curiosity, and I’ll see you there!