Experimenting with Basic Machine Learning “Clustering” When clustering we "experiment" by creating clusters changing at least these four things below.
We can cluster the data by: scaling the data or leaving the data un-scaled (as the rubric suggest, you should try to scale the data at least once to see if this seems to matter)
we can include only 1 column of data when we create a k-means model but we can of course also include 2, 3, 4 or n columns. We could even perhaps add new columns - e.g. poly (column2) or (col1-col3), etc.
if it seems like it may have interesting results. we can change the value of k (of course) e.g. k=1, k=2, k=3 we can plot many things and colour the data points by cluster even if the clustering model was only created with 1 column of data or columns of data that are not being plotted (ggpairs can be quite good with regards to spotting 2-D plots that may have something worth commenting on and will save you from plotting all data pairs.
Once you use ggpairs you can then "zoom in" and just plot 1 or 2 of those individual plots that look the most interesting and/or are interpretation-worthy)
Assignment- 5 (Experimenting with Basic Machine Learning “Clustering”)
I have not attached the dataset but I want you to work on any dataset which is related to “Covid – 19 in Canada” please be sure that the dataset is not Complicated plus I have attached couple of PDF’s from my course with this document, please make sure that the codes used in “R.code” are exactly relevant to the PDF documents from the course and not out of the syllabus or anything new. For reference of the model, I have also attached a sample assignment. I am an undergraduate student so please make sure that the standards are only in those criteria and not more advanced, please don’t make the assignment too complex or complicated by having the codes length too long or out of the PDF’s I have attached.
For this assignment, you will explore your data with unsupervised methods (clustering). “Unsupervised” means that we do not have the traditional idea of y = f(x). We just have many x’s and are looking for patterns without any y-values. In a way, you are looking for “y” values that may be relevant but not yet discovered. In our lecture examples, we came up with the idea of 2 y-values for the faithful data set (long eruptions or short eruptions) and 5 y’s for our malls data set (5 different sorts of customers.
Some suggestions on content:
• Explore 1-dimensional, 2-dimensional, and various dimensional clustering techniques
• Try and make sense of the clusters to see if they have any real-life interpretation from a common-sense point of view
• Use k-means and hierarchical clustering to see if one has any insights the other may not have (or the KNN – K Nearest Neighbours approach)
• Realize that initial conditions of the clustering algorithms matter and therefore perform several runs to see if the initial conditions change the results
• Include your R studio Script as an appendix to your submission
Grading Criteria Is Listed Below, Please DO match the requirements accordingly
Criteria 0 1 2 3
Sufficiency in Exploration Student produced very few insights.
There is no evidence the student attempted very many different x’s and insights are of little value Student has done just “enough.” The work is good, but they scratch at the surface of good ideas they could have pursued further but didn’t The Student has shown enough work to illustrate that they have done sufficient exploring of several variables and several values for K; insights and models are interesting and valuable
Models: the technical workOnly 1 or 2 categories on the far right completed at all, or all 3 completed insufficiently well. 2 or 3 categories to the right completed only reasonably well2 of the categories described to the right done quite well
1. scaling was used appropriately
2.The tools, k-means clustering or hierarchical clustering were used properly
3. various trials under different initial conditions were used to see how that affected clusters
Models: their interpretation Claims made about the meaning of the clusters have great errors in logic or understanding Interpretations are incorrect in a significant way Interpretations are imprecise but generally belivable Student made realistic interpretations of their created clusters; the suggested implications of their models seem logical/ believable
DescriptionIn this final assignment, the students will demonstrate their ability to apply two ma
Path finding involves finding a path from A to B. Typically we want the path to have certain properties,such as being the shortest or to avoid going t
Develop a program to emulate a purchase transaction at a retail store. Thisprogram will have two classes, a LineItem class and a Transaction class. Th
1 Project 1 Introduction - the SeaPort Project series For this set of projects for the course, we wish to simulate some of the aspects of a number of
1 Project 2 Introduction - the SeaPort Project series For this set of projects for the course, we wish to simulate some of the aspects of a number of