Dossier Analyse de données - Statistiques - Data file analysis (en anglais) - Meilleure note obtenue.

Thèmes abordés

Analyse de donnée, statistique, dossier, chi2, Khi2, χ2, tableau, anglais, english, data file analysis, statistics, variable, plot, diagramme, chart, regression, table, statgraphics

Résumé du document

Dossier réalisé dans le cadre d'un cours d'analyse de données - statistiques en Master.

Logiciel utilisé : STATGRAPHICS Plus.

MEILLEURE NOTE OBTENUE


Titre : Mercury in Largemouth Bass in Florida Lakes


Ce dossier, en anglais, analyse donc un tableau de données récoltés dans les différents lacs de Floride (alcalinité, pH, calcium, chlorophylle, mercure).

Cette analyse a pour but de trouver des relations entre les différentes données du tableau.

Le sujet peut paraitre étrange mais il s'agit là d'un acte délibéré afin de produire un dossier original.

Extraits

[...] Finally, we noticed on several occasions in our analysis that the extreme values observed during all of our tests mainly came from the same lacs numbered and 45. Those lacks are Hamey, Puzzle and Talquin but it is impossible for us to explain those outstanding values by any geographic and/or geological exposition. Nature has it's secrets . [...]

[...] Thus, in a first part, we will show some one-variable statistics to have a better and a clearer approach of the table and all its components, and by this way a better control of the situation. In a second hand, we will analyse the different possible links between each variable in order to have a better comprehension of the degrees of influences existing in all these relations. Of course, we will point out very attentively these influences toward the average mercury. [...]

[...] If we want to classify these variables by order of influence towards the Avg_Mercury, Chlorophyll will arrive first (confirmed by the highest values of this variable for which ones the Average of Mercury is nearly non- existent). Then, it will be Alkalinity and Calcium (we can easily notice the highest values of these two variables but they have not an important impact on the mercury rate, and they can be classified as outside points). We could represent these 3 variables with a decreasing exponential curve. Concerning the pH, it seems that it has a very different influence. It could be represented by a decreasing linear curve. [...]

[...] Indeed, with the median, all the counts would have been the same. This share is represented in the table below. Then, we applied the test on this table Count table ( Theoretical count / Count ) 4.2 Contingency Tables Those tables show the results of a hypothesis test run in order to determine whether or not to reject the independence between the row and column classifications. If the P-value is greater than or equal to 0,05, we cannot reject the hypothesis that rows and columns are independent at the 95,0% confidence level, that is to say if this rate of 0,05 is completed, we can claim with assurance that rows and columns studied are dependent of each other in a certain degree. [...]

[...] Since the P-value is less than 0,05, we can reject the hypothesis that rows and columns are independent at the 95,0% confidence level. As the P-Value is equal to 0,0068, that is to say below the limit, there is definitely a link between the 4 variables. It confirmed the synergy observed previously and the relations in the behaviour of each variables toward the others. Indeed, Alkalinity, Calcium and Chlorophyll have similar behaviour whereas pH acts in a completely opposite way. [...]