Daily Data Science Briefing — Monday, August 17, 2026

🧠 Data Science Interview Practice

Topic Category: Machine Learning Fundamentals (rotation: day-of-month 17 mod 8 = 1)

Question: "Walk me through the bias-variance tradeoff. Given a model that's performing poorly, how would you systematically diagnose whether the problem is bias or variance, and what would you do about each?"

Model answer:

• Definitions: Bias is error from overly simplistic assumptions (model can't capture the true relationship); variance is error from sensitivity to fluctuations in the training set (model overfits noise). Total expected error ≈ Bias² + Variance + Irreducible error.
• Diagnosis via learning curves: Plot training error and validation error against training set size.
  • High bias: both training and validation error are high and converge to a similar (poor) value — more data won't help much.
  • High variance: training error is low but validation error is much higher, with a persistent gap — more data or regularization helps.
• Quick heuristic: compare training error to a baseline/human-level error, and compare validation error to training error.
  • Train error >> baseline → bias problem.
  • Val error >> train error → variance problem.
• Remediation for high bias: increase model capacity (more features, deeper/wider model, less regularization), engineer better features, reduce regularization strength, train longer.
• Remediation for high variance: get more training data, add regularization (L1/L2, dropout), reduce model complexity, use ensembling (bagging), or apply early stopping.
• Code snippet (learning curve check in sklearn):

from sklearn.model_selection import learning_curve
import numpy as np

train_sizes, train_scores, val_scores = learning_curve(
    estimator=model, X=X, y=y, cv=5,
    train_sizes=np.linspace(0.1, 1.0, 5),
    scoring='neg_mean_squared_error'
)
train_mean = -train_scores.mean(axis=1)
val_mean = -val_scores.mean(axis=1)
# Large, persistent gap -> variance; both high & converged -> bias

Follow-up an interviewer might ask: "If you have high variance and can't collect more data, what are three techniques you'd try, and how would you decide between them given limited compute budget?"

🇰🇷 Korea Morning News

1. Trump orders reduction of US–South Korea military drills. President Trump directed the Pentagon to scale back joint exercises with South Korea, including the ongoing 11-day Ulchi Freedom Shield drills (Aug 17–27, ~18,000 troops), citing his relationship with Kim Jong Un and calling the drills costly and "hostile" toward the North. — Al Jazeera

2. North Korea condemns joint drills as provocation. Pyongyang's Foreign Ministry criticized US–Japan–South Korea military cooperation as edging toward a "nuclear alliance," warning it would respond with a "new level of deterrent." This follows a North Korean ballistic missile launch earlier in the week. — Al Jazeera

3. Seoul eases rules to unlock ~4.3 trillion won ($3B) in investment. The government announced regulatory and permitting changes at an emergency economic meeting (chaired by Deputy PM/Finance Minister Koo Yun-cheol) to accelerate investment in semiconductors, batteries, and biotech. — UPI

4. Korea unveils 7 tech projects for future growth engines. The government announced initiatives spanning renewable energy, space science, and other advanced technologies as part of a broader push to diversify growth drivers beyond semiconductors. — The Korea Times

5. 2026 GDP growth forecast raised to 3.3%. Economists (Bloomberg survey) lifted their South Korea growth outlook from 2.8% (July) to 3.3%, driven by strong global demand for semiconductors and a record trade surplus. — Bloomberg

📈 US Stock Market Briefing

Sentiment: Mixed-to-cautiously-positive. Futures wavered Monday morning: Dow futures -0.2%, S&P 500 futures +0.1% (after a third straight weekly gain), Nasdaq-100 futures +0.5%. The S&P 500 hit a fresh record high this past week.

Notable movers/sectors:

• Tech and AI-adjacent names led premarket gains, fueled by strong revenue growth reported at Anthropic, reinforcing confidence that AI infrastructure spending will continue.
• Semiconductor/storage names (Sandisk, Micron) rallied in premarket trading.
• Retail sector in focus ahead of a heavy earnings slate this week (Walmart, Target, Home Depot, Lowe's, TJX).

Macro factors: Investors are watching the Fed's next policy move; recent soft inflation data helped drive last week's rally. No major economic releases are scheduled for today specifically — attention shifts to this week's retail earnings and oil prices, with a lighter macro-data calendar than recent weeks.

Sources:

• Trump says US to 'substantially reduce' military drills with South Korea — Al Jazeera
• N Korea launches ballistic missile as S Korea, US plan military drills — Al Jazeera
• South Korea eases rules to unlock $3 billion in advanced investment — UPI
• Korea unveils 7 tech projects for securing future growth engines — The Korea Times
• South Korea Growth Forecast Raised to 3.3% for 2026 on Chip Export Surge — Bloomberg
• Stock market today: Dow, S&P 500, Nasdaq futures waver ahead of retail earnings — Yahoo Finance
• Stock market next week: Outlook for Aug. 17-21, 2026 — CNBC