Choosing the right kind of model
for machine learning
Imagine that you are a carpenter. What kind of tools do you think you will
have loaded in your truck when you arrive at a worksite? You will probably
have a hammer, and a drill as well as a few different types of saws. You
probably have a few planes and a good set of drill bits. If you know how to
do your job, then you know when you will know the purpose of each of
these tools, and you will know when each one should be used. Every single
one of these tools serves a specific purpose. The drill can’t do a hammer’s
job, nor would you try to cut something with a hammer.
A data scientist who is looking to do machine learning will have his own set
of tools, each serving a different purpose and designed for a different
function. Depending on what kind of data you’re using, and what you want
to find out, you will have to choose different algorithmic models to do the
job.
Statistical algorithms can serve several purposes. Some predict a value; like
a regression model that predicts your income base on your years of
education and work experience. Some models predict the likelihood of an
event occurring, like a medical model that predicts the likelihood of a
patient surviving a year, or two years, etc. Other models sort things by
placing them into different categories or classes, like a photo recognition
software sorting photo of different types of dogs.
Depending on the outcome you are looking for, you will need to have your
statistical toolbelt. You need to familiarize yourself with the technical skills
of statistics. Also, you will have to know which tool you want to use, and
when to use it. Here I have created a comprehensive list of the different
kinds of statistical models that are very common in machine learning. To be
able to write the code to build these models on your own, I recommend that
you take some time to study in the programming language that you've
chosen. But this list will give you an introductory understanding of each
type of model, and when they are useful.
For machine learning to be effective, you must choose the right model and
the model that works best and have relevant data for the model and the
question at hand.
Nowadays, especially with the use of the internet and digital marketing,
there are certain questions that can’t be properly understood without the use
of data and machine learning that can analyze it. With machine learning and
data science, you can keep track of your customers and their buying habits,
so that you can better adapt to their needs when they change.
The better you are at interpreting your data, the more easily you will be able
to identify trends and patterns so that you can anticipate the next change.
Machine learning can be broken down into three different categories, each
one containing several unique algorithms that serve different purposes. To
start, we’ll talk about the differences between supervised, unsupervised, and
reinforcement learning.
Do'stlaringiz bilan baham: |