Edited by David Pennington, Friday, 29 Jul 2016, 17:14
I can't believe that we are almost there. The EMA is due in early September and it is now the end of July. In my last post I was explaining that I was working at getting just enough to pass both TMAs with a 40% average. I started on the EMA and had a major issue (about using my own code to do the work rather than Python/Pandas) and was worried about my TMA mark. My tutor had to go away so, even though I got the work in early, it was going to be some time before I got the result. I was talking to her about how to approach the TMA with my coding issues when she threw in. - well your recent work has been very good!
When I got the TMA back, amazingly I had got 77% so I even beat the first one. Wow. Maybe I did understand more than I thought. My tutor advised me that, although I could use whatever I wanted to complete the EMA, I should only expect the marker to understand the tools in the course. She also advised me that I should choose two reasonable questions but it is the nature of this type of data that any question may prove to be unanswerable.
I spent many days investigating the bat data using both Excel and my own code and slowly formed some questions that looked reasonable. Everyone seems to be having a difficulty in incorporating the k-nn or k-means calculations into the final report but, suddenly one night around 1am when I woke up, I could see a use of k-means.
Finally, in the last tutorial, I also became aware of another aspect of all of this. This EMA isn't about coding. It is about analysis and reporting and, although there is a big weighting on the techniques, this doesn't actually mean that I have to write loads of code - just use the tools we have been given (which includes Excel) and "write a good report". The bulk of the marks are got by this and my tutor has given me some really great advice - both in the tutorial and in my TMA.
Incidentally, I got 60% in the iCMA for answering only four of the eight question. I did this by re-writing the code in Smalltalk. I now have coded k-nn, k-means and Cosine Distance which means that I have a great handle on what these are doing.
Final word. I am getting the data sorted and getting the nuts and bolts of the report under way. I can't see me looking at the last units of this course until I have submitted the EMA. I am interested in the "Semantic Web" but, as it isn't tested in the EMA, it can wait. I don't even need to do any of the last iCMA.
TMA02 marked and now only the EMA to go
I can't believe that we are almost there. The EMA is due in early September and it is now the end of July. In my last post I was explaining that I was working at getting just enough to pass both TMAs with a 40% average. I started on the EMA and had a major issue (about using my own code to do the work rather than Python/Pandas) and was worried about my TMA mark. My tutor had to go away so, even though I got the work in early, it was going to be some time before I got the result. I was talking to her about how to approach the TMA with my coding issues when she threw in. - well your recent work has been very good!
When I got the TMA back, amazingly I had got 77% so I even beat the first one. Wow. Maybe I did understand more than I thought. My tutor advised me that, although I could use whatever I wanted to complete the EMA, I should only expect the marker to understand the tools in the course. She also advised me that I should choose two reasonable questions but it is the nature of this type of data that any question may prove to be unanswerable.
I spent many days investigating the bat data using both Excel and my own code and slowly formed some questions that looked reasonable. Everyone seems to be having a difficulty in incorporating the k-nn or k-means calculations into the final report but, suddenly one night around 1am when I woke up, I could see a use of k-means.
Finally, in the last tutorial, I also became aware of another aspect of all of this. This EMA isn't about coding. It is about analysis and reporting and, although there is a big weighting on the techniques, this doesn't actually mean that I have to write loads of code - just use the tools we have been given (which includes Excel) and "write a good report". The bulk of the marks are got by this and my tutor has given me some really great advice - both in the tutorial and in my TMA.
Incidentally, I got 60% in the iCMA for answering only four of the eight question. I did this by re-writing the code in Smalltalk. I now have coded k-nn, k-means and Cosine Distance which means that I have a great handle on what these are doing.
Final word. I am getting the data sorted and getting the nuts and bolts of the report under way. I can't see me looking at the last units of this course until I have submitted the EMA. I am interested in the "Semantic Web" but, as it isn't tested in the EMA, it can wait. I don't even need to do any of the last iCMA.