shivtosh / feature-engineering Goto Github PK

This file consists of the .csv file with raw data picked up from kaggle.
The data is cleaned using R (rmd file).
file_for_tableau.csv represents dataset cleaned and ready for tableau.
The final .twbx file is visualisation made with tableau. It also includes the dashboard prepared using the sheets in tableau.
The report guiding with the steps of the project and insights captured is uploaded in pdf format.

diabetes in a patient.

INSIGHTS

• Interesting HDI trends are observed between diabetic and non-diabetic population.
• Basically diagnosis of diabetes points towards a lower HDI.
• Self care practices are more prevalent in diabetics than non-diabetics. Males are more vigilant than females.
• Lifestyle choices of heavy alcohol consumption or smoking do not point towards a substantial correlation. In fact whether diabetic or non diabetic, more people opt for a healthier lifestyle than indulging themselves abusively.

CRITICAL REFLECTIONS

• The population of diabetics and non-diabetics was not balanced and hence comparison between the two could only be done on a proportion of the total basis.
• Lack of exact numerical data and presence of interval data instead (for income, education and age) is a bit inconvenient for the calculation of dimension indices and for final calculation of Human Development Index (HDI). This kind of calculation can be done in Tableau too.
• The above is not just inconvenient but also more inaccurate than it would have been with exact data.
• HDI is an indicator of life expectancy and standard of living and not of inequality or poverty and hence its correlation with diabetes diagnosis does not imply any lack of financial resources.