Đại học Kinh tế Quốc dân – National Economics UniversityMICRO4ASIA
🏆 Arena
Sign inSign up
MICRO4ASIAThe six micro-credentials

Courses are delivered on the Limio learning platform. Limio

← Mathematical Foundations for Data Science

You are viewing a public preview

Sign in to save your progress, do assignments, join discussions and get feedback.

Sign in
Part III — Probability and Statistics

Topic 15 · Statistical Estimation and Confidence Intervals

Print / Save as PDF
Lecture video — Topic 15: Statistical Estimation and Confidence Intervals

Parameters and estimators, bias and mean squared error, confidence intervals for a proportion and a mean, efficiency and consistency, the method of moments and maximum likelihood.

Flashcards · Estimation and Confidence Intervals

10 words
WordMeaningExample
  • Parameter, statistic, estimator, estimate

    A parameter is a fixed population quantity; a statistic is a function of the sample; an estimator is a statistic used for a parameter (random); an estimate is its observed value.

    p, K, p̂ = K/n, and 40/200 = 0.20.

  • Bias

    B(θ̂) = E(θ̂) − θ. Unbiased means E(θ̂) = θ for every parameter value in the model.

  • Mean square error

    MSE(θ̂) = E[(θ̂ − θ)²] = V(θ̂) + [B(θ̂)]².

    A: 0.04 + 0 = 0.04; B: 0.02 + 0.10² = 0.03.

  • Sample variance

    S² = Σ(Xi − X̄)² / (n − 1), unbiased for σ². The normal MLE uses n and is biased by the factor (n − 1)/n.

    Data 1, 3, 5: s² = 8/2 = 4.

  • Confidence interval for a proportion

    p̂ ± z·√(p̂(1 − p̂)/n), with z = 1.96 for 95%; approximate.

    40/200: [0.145, 0.255].

  • Confidence interval for a normal mean

    x̄ ± t·s/√n with t from the t distribution with n − 1 degrees of freedom.

    n = 16, x̄ = 10, s = 4: [7.869, 12.131].

  • Meaning of confidence level

    The long-run fraction of intervals, built by the procedure over repeated samples, that contain the fixed parameter.

  • Efficiency and consistency

    eff(θ̂1, θ̂2) = V(θ̂2) / V(θ̂1) for unbiased estimators; above 1 favours θ̂1. θ̂n is consistent if P(|θ̂n − θ| > ε) → 0 as n → ∞.

    Variances 4 and 9: eff = 9/4.

  • Method of moments

    Set population moments (functions of the parameters) equal to sample moments and solve.

    Uniform(0, θ): θ/2 = x̄, so θ̂ = 2x̄.

  • Maximum likelihood

    Choose the parameter that maximizes the likelihood of the observed data; usually maximize the log-likelihood.

    8 successes, 12 failures: p̂ = 8/20 = 0.40.

Practice in Python

Work through Section 6 of Lab 3 (Probability and Statistics, MFDS_Lab3_Probability_Statistics.ipynb), attached to the Part III materials lesson at the start of this part.