|
sungnr: I am afraid I am not able to do so. As ETS is a profit taking business entity, it must at least have annual schedule about database change. Under the assumption they have annual plan rather than, say, five years, unless I have at least two years data to study the pattern, any results would be unreliable. If you have enough data, say you tracked when they change the database for more than two years. One logical method would be perform the descriptive statistics on the number of days per KFC. Find out the median, mode, range, maximum, and minimum. And you can use histogram to study the frequence for each sample points. Also, you can compare data sets between two years. In CFA, they call datamining a way to 'Beat the data into submission'. God I love this saying.... For the data set given above, both 2007 and the first 8 months of 2008, Days per KFC |
|
| Mean | 31.11111 | Standard Error | 0.956194 | Median | 30 | Mode | 30 | Standard Deviation | 4.056786 | Sample Variance | 16.45752 | Kurtosis | 1.767511 | Skewness | -0.10832 | Range | 17 | Minimum | 21 | Maximum | 38 | Sum | 560 | Count | 18 | Confidence Level(95.0%) | 2.017392 |
Days | Frequency | Cumulative % | 0 - 20 | 0 | 0.00% | 21 - 25 | 1 | 5.56% | 26 - 30 | 9 | 55.56% | 31 - 35 | 5 | 83.33% | 36 - 40 | 3 | 100.00% | > 40 | 0 | 100.00% |
You figure out the rest...... | 
京公网安备11010202008513号 京ICP证101109号 京ICP备12012021号
发表于 2008-8-7 00:41:00
