ISSS608 2017-18 T3 Assign Alejandro Llorens Moreno Conclusion
VAST Challenge 2018:Like a duck to water
|
|
|
|
|
Conclusion
Top 3 patterns
1) We have identified several challenges in the sampling strategy. Locations and Chemicals have very different distributions which seem quite suspicious. Why taking so many samples of a specific location and few from another? This is worth investigating further.
2) Specific outliers (and extreme outliers 3 standard deviations from the average of the samples taken) are present in this dataset. We have presented a few examples worth investigating further. For example, Methylosmoline, Nitrates, Magnesium, Sulphates, Coliforms, Iron..etc.. are just a few examples of samples that represent very high values in specific months. Understanding the relationship between these variables and correlating these with the outliers presented in our Tableau analysis is important to interlink suspicious potential behaviours.
3) Designing a better sampling strategy is important. A strategy more systematic and focusing at specific days of the month over time. This would help determine cyclical patterns or spot true outliers in the data. Most of the chemical values have been discarded because of poor quality of data, not enough samples or abnormal distributions.
