My data has only 3 columns. These columns are year_2012, year_2010, year_2007. These columns are about measurements performed in various years (2012, 2010, 2007). Rows are countries (~150 countries/rows).
Measurements are all numeric and any value between 1.00 through 5.00. 5 is the best, 1 is the worst.
For instance, how can I extract a group of countries with better/worse progresses from year 2007 to 2012? What else can be done on such data?
>For instance, how can I extract a group of countries with better/worse progresses from year 2007 to 2012?
By simply looking at the values? No, honestly, read the manual of the software you are using.
>What else can be done on such data?
Strange question. Why do you have this data? Why was it collected? Why do you want to analyse it? There should be some purpose. Actually, FIRST the purpose is clear, there is some specific question. Then the experiment or data collection is planned so that the data (will) best serve the question. According to the kind of data and the question, it is anyway clear how the data has to be analyzed to best answer the question.
Dear Jochen, the data are about logistics performances of countries. My purpose is to see the "progresses". For instance, which country or countries have higher moments or positive "trends" through better performance, from year 2007 to year 2012.
For example, I might calculate a linear trend line for each row and cluster them based on the angle of trend lines to see progresses. Or, try to predict row values for year 2014. Any more suggestions?
I see. Exploratory analysis. Ok. Clustering is fine. You can try different distance metrics (also correlation, what comes closest to your "linear-trend-line-clustering"). But prediction is something entirely different. It is quite an over-optimistic assumption that such data would provide a basis for useful predictions!
- Badges
- Science topic