Time: 2 weeks

Team: 2

Language: Python

Several months before an important election, many polls seem to pop up from nowhere. Their interpretations are often surrounded by uncertainty: to what extent are these polls reliable? Why are there so many differences between poll institutes? And from day to day? Is a 3% variation significant? etc...

To estimate the accuracy of the results, a confidence interval is given. It is defined by the fact that there is a x% probability that this interval encompasses the true value.

You already know that questioning people follows a Bernoulli process, and therefore that a binomial distribution (converging toward a normal distribution) is a good model for the results. You can then easily compute the confidence intervals, knowing that:

• The 95% confidence interval amplitude is 2 × 1.96√v
• The 99% confidence interval amplitude is 2×2.58√v

where v stands for the variance of the sample proportion (corrected for finite populations).

The goal of this project is to compute the 95% and 99% confidence intervals.

>> ./209poll -h
USAGE
    ./209poll pSize sSize p

DESCRIPTION
    pSize   size of the populations
    Size    size of the sample (supposed to be representative)
    p       percentage of voting intentions for a specific candidate

Author Corentin COUTRET-ROZET and PATRICIA MONFA-MATAS