0
$\begingroup$

I hope this message will find you well.

First of all, I'd like to apologize for not using the most commonly used words in statistic's field : I have no background at all in that, and my knowledge is limited to memories from math lessons, and some thoughts I got reading articles these past weeks. Most of them would point that it might be necessary to use dummies, or strongly advise for. (like that old one)

I would like to avoid that, due to the extra processing (I would end in having so many columns) and increased difficulty to use the result. Althought, I will have to do it once, in order to look for correlations between them and the outcome, in order to reduce the weight of datas. But once only, as i will have to re-use the algorithm on a regular basis, with the original data set structure, but new datas. I will check once in a while if there is no change in most important factors, but it should not as the environment is destined to evolve really slowly.

My data set has no missing value at all. Most of the variable are categorical (non ordinal), some are discrete. My dependent variable (the outcome I'd like to predict) is one of those discret variable.

I followed some Python tutorial and I'd like to stick to that language (if possible) in order to increase my knowledge and understanding of it.

So my question is : is there a library/module in Python I should read the documentation from in order to be able to predict my outcome according to my data structure ?

Do not hesitate to let me know if I'm not clear enough (said I was far from being a pro !), I would gladly edit my message.

Thank you for reading, Thank you even more for answering !

have a nice day,

EDIT following jdjame's answer To explain the context a little bit more, as it looks I was not specific enough. (as expected according to my initial message !) My data set regroups different characteristics of employee :gender, age, seniority, type of work, type of contract, affectation, and so on, and number of planned working days with all those characteristics and number of days of absence on planned working days with all those characteristics

My goal is, using previous behavior, to predict the number of absence days to expect in the next month on a specific job/location. I do not need specific visualization (expect for the first time to determine which determinants I should extract for my training base), as I would rather export the results to Excel-like file. (Might change idea on that point later on, as I learn, but it is not a priority) Every month, I would enrich my data set with the last ended month (and probably delete the oldest data so computer would still accept to run without burning (ie freezing)), so I can get more accurate / actual results.

I hope this edit is useful to you to provide me a more precise answer, even if our previous one already gives me some tracks to follow !

Thank you again.

$\endgroup$
1
  • 2
    $\begingroup$ I'd be shocked if you couldn't use scikit-learn. $\endgroup$ Commented Jun 15, 2020 at 20:46

1 Answer 1

1
$\begingroup$

There are quite a few available python libraries that could help you work with data. If you could be more specific on the task you wish to accomplish that would help in supplying a better response. Right off the bat, I would suggest using the pandas library. You can look into the documentation here: https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.html. It will help you store your data into a DataFrame object (essentially a table) that will be more tractable. From there, it is a matter of what you wish to do with your data. To visualize it, I would suggest matplotlib which is a great library for graphing points (https://matplotlib.org/). Depending on the classifier you wish to use, you may want to use an sklearn supervised learning module (https://scikit-learn.org/stable/supervised_learning.html). If you wish to make a deep neural network, then I would suggest starting out with keras (https://keras.io/getting_started/) or pytorch (https://pytorch.org/get-started/locally/). It is important to note that all of these libraries are meant to work with machine learning data so many fo the preprocessing steps you might want to do (i.e. processing your categorical values). Hopefully this provides a helpful start for you.

$\endgroup$
1
  • $\begingroup$ Thank you a lot for your time. I did edit my initial message to give some more insight on my task, I hope it will be helpful to you for helping me ! I did vote for your answer, but haven't enough reputation for it to be counted. $\endgroup$
    – Jacques
    Commented Jun 15, 2020 at 21:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.