Sie sind auf Seite 1von 3

Just as the picture suggests, a Binary Tree is a tree data like structure which elements
of more than one or two nodes were contained. Each nodes represents a feature or
type connected with various branch representing as a decision to come, thus making it
off similar as a decision tree. The whole idea for making this tree-like structure is for the
entire data and processes to eventually reach its’ outcome or a possible result at the
very end. Machine learning also works by this concept of Decision tree method. By
analyze the data given, converting them to numbers for the AI to train, and try to predict
the outcome of the data it fits into, until finally prints out the results for the outcome is
also part of machine learning program as well.
The topic that I decided to create while trying to simulate the machine learning program is about
gemstones and each of its families, like garnet and quartz for example. The objective of writing
the code is to try and train the AI in order for it to be able to predict and seperate all of gemstone
families as much as possible. Each family has a different variety of gravity weight, the moh scale
of hardness, its index of refraction, as well as different colors added to them as well. The data
that was collected is also according to the real world’s data, as shown in the link here:
https://www.gemrockauctions.com/learn/technical-information-on-gemstones/mohs-hardness-sc
ale-for-gemstones

However, I also faced with mistakes while trying to come up with the data as well. Because
three of the four datas that I used needs to tell the number at the exact amount, I need to
measure them correctly else my code will be messed up. I also need to list the correct colors for
each of the stone’s family as well so I decided to make a guide code in case that I wouldn’t
mixed the colors up with other unrelated data as well.
My AI code proves to be accurate at a reasonable enough, with the AI able to split each
requirements for the gemstones like gravity, hardness, and index of refraction usually correct.
However, while the result works for gems like Garnet and Quartz. Tourmaline and Chrysoberyl
seems to have conflict over the colors, as one of the colors that belongs to the Tourmaline
family actually shows up as Garnet instead. Which means that the AI predicted by only the
numbers of the three data mentioned above, but the colors were left off. Which indicates that
there might be a possibility of messed up numbers in my final test data chart. Other than that,
the AI seems to be able to predict data which includes all the gemstones, which is fine for me.

Das könnte Ihnen auch gefallen