The results of the competition may be not very good, there have been a series of one way or another problem, this time there are so many ideas to write this, as a reminder to himself after
The biggest mistake first game is timed out, why overtime, I thought for a moment:
1, category and are not familiar with the contents of a mathematical model, this time the problem is that this type of model we have never in contact with the evaluation team and multivariate analysis, cluster analysis can be classified not know, I do not know discriminant analysis, do not know European, Mahalanobis distance discrimination law, know, and will not use, all in the game to know more, I think it is the most pit fisher discrimination law, I was going to use, and as a method to separate healthy and unhealthy, but his the premise is established in the case of the topic or they have to separate categories of data belonging to the new method to determine what kind of team we put it to abandon, at a time when two-thirds of the time has elapsed, another model of abdominal fetal death in result, the learned comments about the process, the teacher data by a first examination the patient asked analysis, cluster analysis i.e. the separation of the categories and extract characterized as basis for evaluation, and then using the second configuration discriminant asked the drawbacks of data classification, compared to the entropy law by seeking to further build the right criteria for the much more appropriate, and to the evaluation system entropy law is that if the number of According to the obvious heavy weights without bias, the size of the data played a key role, which leads to the second question when using gray correlation method, the correlation even the most standard values are not big big data correlation, so the evaluation system complete collapse.
2, the blind use of model
This is what impressed me most is the principal component analysis, just think the main factors that can find out, you really feel that we can go and programming, compile complete discovery, index weight is unknown, the main component is a composite indicator, which is the so-called dimensionality reduction, not so easy to extract easy to go back, blind use of time-consuming model is the culprit
3, team communication
I think sometimes team communication is not very good, sometimes I was busy programming, problem-solving ideas but need to write the entire problem-solving program, I do not want my train of thought was interrupted programming, so do not want to say, sometimes drawing in Paint is programmed not want to be disturbed, each re-think, and in order for a long time, but the whole thesis which leads out of line with my thoughts, because writing was the main person is writing, if he did not think Mo was equal to the entire team think, will be very low, the article did not express the whole team really thinking, and I personally belong to the kind of difficulties will be impatient, can be affected by my dad (he was back pot), to be honest king was profound temper can be good, no pressure to communicate with him, may he a lot of pressure now.
4, drawing
For a long time did not go to learn the drawing, the most impressive is a histogram, really autistic, if not clear characteristics and scope of each plot, then draw the diagram was a disaster, the histogram for statistics similar numbers of the number and frequency, is not suitable for the expression of a series of time-series data has characteristics, as a percentage or scatter plots may be more appropriate, in short, still need to learn, mainly take the time to spend less