3
Hello I have some data that I tabulate in excel monthly, which is very repetitive. I wonder if there is any way to use machine learning in R to teach an algorithm, based on the data I already have, fill in the new data that arrive. Basically I receive the data in columns with ID, origin, description, quantities, values and etc. About this data I tabulate the detailed instructions that I need, as for example the brand (ex: I look at the description column and check which is the brand) and so successively I put in other columns in the same spreadsheet the tabulated data of the desired form. This process is soon too repetitive and therefore takes a lot of time. Reading about Machine Learning I saw that in R language I can train an algorithm to analyze my data tabulated previously and based on that the algorithm could provide me what would be the brands, producers, origins and etc.
My question is, how can I apply this to my work using R to speed up the process?
Ps. If it is not yet clear forgive me for it is really difficult to describe.
Yes, there are several possible ways. Did this "answer" help you? No? : ) It is... it is the same as your question: broad and generic. Edit the question to improve it, trying to describe your problem with practical examples. Otherwise, no one will be able to help you. Also focus on the problem by posting any R code you already have. If your question is very "what is the best ML technique", it may not be part of the scope of this site (and then, it would be better to post on Cross-Validated). Ah, and if you haven’t done it yet, do the [tour] and read [Ask].
– Luiz Vieira
Hello, I tried to improve!! Sorry.
– Bene
It is still wide and is now less clear. You say that you fill columns based on other columns looking at the mark (for example), but this has nothing to learn, just observation. Any data extraction algorithm does this. If the work is repetitive, you need automation and not machine learning. You look like you need VBA, not of R.
– Luiz Vieira
I suggest reading other questions here. Start with the tag [tag:vba], for example. It will give you an overview of what you can do. If you still think you really need ML, read the questions on the tag [tag:machine learning]. Particularly this question can be useful for an overview.
– Luiz Vieira
P.S.: Machine Learning will be useful to you if you wish foresee a class (the brand of a product, for example) from some data set which correlates with the class. Note that this will not necessarily be true for any data. For example: while it may be possible to predict the brand of a product based on user reviews on a website or the number of sales (because there may be a relationship due to differences in quality), it is unlikely that the brand can be predicted based on weight (since the same product in different brands can).
– Luiz Vieira
The field I use to extract the Infos is the description field, which based on a history I have, I can know that that description is of such a mark.
– Bene
If you are really going to use ML, you will have a lot to learn before you are able to have a solution. Since the relevant data are not numerical, but text, it needs to have an intermediate step of characteristic extraction. You’ll get an idea of this at this link. So, if it is possible to do a simpler automation with VBA, I think you should go that way, okay? Good luck.
– Luiz Vieira