# Confidence Interval Filter

## Contents

### Overview

Confidence intervals consist of an interval of values that is assumed to be a good enough estimator of an observed data set. The level of confidence of such interval indicates the probability that the confidence range captures the observed value within a distribution of samples. Basically, a 95% confidence interval means that we are 95% confident that a true new value of the observed fata set belongs to our confidence interval. In many practical applications confidence intervals are typically stated at the 95% confidence level. However, other commonly used confidence levels are 50%, 95% and 99%.

### Practical example

A machine fills bags with 250 grams of rice. Clearly the machine cannot fill every bag with the exact quantity of rice and the actual content is considered have random fluctuations around the desired average and desired of 250 g. To determine if the machine is adequately calibrated, a sample of 10bagsis chosen at random and weighed. Based on these measurements we found an average weight of 250.7 grams. Should we consider this computed value statistically unusual, i.e. statistically outside our confidence interval of 95%? In this specific case, assuming the machine fills bags with a normal distribution of standard deviation of 2.5, we will find that the 95% confidence interval is between 249.2 and 251.2, this leading to the conclusion that the observed measure of 250.7 is within the interval an thus statistically within our designed confidence interval.

### Computing a the Confidence Interval

Computing the Confidence Interval is not straightforward as it required to compute the values of the Cumulative Distribution Function for the given probability endpoints via an iterative process:

### Within Grapheme

To create a Confidence Interval Filter within Grapheme, click on Manage Views icon in the Table Editor toolbar and Edit the Current View. Add a new Probability Interval, choose the columns that will be used to evaluate the filter and insert the Confidence Interval value (%).