91Ó°ÊÓ

Maximum Likelihood Estimation, commonly referred to as MLE, is a fundamental method used in statistics to estimate the parameters of a probability distribution. For instance, when dealing with a set of data presumably drawn from a normal distribution, you may want to determine the mean and variance that best fit this data.

MLE works by maximizing a likelihood function so that, given the constraints of your model, the probability of obtaining the observed data is as high as possible. In simple terms, MLE attempts to "explain" the data in the best way through the parameters. In our specific context of the exercise, we are interested in the mean, \(\mu\), of a normal distribution.

• Under the Null Hypothesis (\(H_0\)): The maximum likelihood estimate for the mean, \(\mu\), is simply the specified value, \(\mu_0\).
• Under the Alternate Hypothesis (\(H_1\)): The MLE for \(\mu\) is the sample mean, denoted as \(\hat{\mu} = \frac{1}{n}\sum_{i=1}^{n}y_i \). This sample mean is the average of all our observations.

Understanding MLE is essential because it provides a systematic approach for parameter estimation, grounding the basis of many statistical tests and models.

The Normal Distribution, often described as a bell curve due to its shape, is one of the most important probability distributions in statistics. Real-world phenomena such as heights, test scores, and measurement errors often follow a pattern that can be represented by a normal distribution.

Key characteristics of the normal distribution include:

• Symmetry: It is perfectly symmetrical about its mean, meaning half the data fall below the mean, and half above.
• Mean, Median, and Mode: In a normal distribution, these three central tendency measures are identical.
• Standard Deviation: The spread of the distribution, with approximately 68% of data falling within one standard deviation of the mean.

In the context of our exercise, the sample is drawn from a normal distribution with a known variance of 1. This simplifies our calculation since, for hypothesis testing and maximum likelihood estimation, we only need to focus on estimating the mean, \(\mu\). Once we have these estimates, we can use them to determine if the mean of our sample conforms to a hypothesized mean.

Hypothesis Testing is a statistical process used to make decisions about the properties of a population, based on a sample. It involves comparing the null hypothesis \(\(H_0\)\) against an alternative hypothesis \(\(H_1\)\).

The null hypothesis typically represents a default position or stated assumption, often suggesting "no effect" or "no difference". In contrast, the alternative hypothesis represents what you aim to support or confirm through your test.

In our exercise, the GLRT (Generalized Likelihood Ratio Test) is employed to test if the mean \(\mu\) is equal to a specific value \(\mu_0\) (\(H_0\)), or not equal to \(\mu_0\) (\(H_1\)).

• The process starts by defining both hypotheses clearly.
• The test statistic is computed, which in this case is \(-2\log(\Lambda) = n(\hat{\mu} - \mu_0)^2\).\
• A higher test statistic value suggests stronger evidence against \(\H_0\).\
• The significance level \(\alpha\) is chosen, which represents the probability of rejecting \(\H_0\) when it is indeed true.

The hypothesis test concludes by deciding whether or not to reject \(\H_0\) based on the calculated test statistic and significance level, ensuring objective decision-making based on the available data.