Projects involving data science

[EngWiPy]

Hello,

I am trying to do some projects on data science/machine learning using Python, but I am not sure what to do. I downloaded a very simple dataset from WHO, and I am trying to do something with it, but most of (actually all) what I can do with it is data manipulation and exploratory data analysis (histograms, scatter plots, ... etc). I need these projects for my CV since I don't have previous experience in the field. Any suggestion will be highly appreciated.

Thanks

 

[jedishrfu]

Do something with acoustics like voice recognition, or animal sounds recognition. There are a lot of datasets for acoustics that could be used.

Image character recognition is a common project. Here's a video on how neural-nets work that uses the character recognition.

 

[EngWiPy]

Do something with acoustics like voice recognition, or animal sounds recognition. There are a lot of datasets for acoustics that could be used.

Image character recognition is a common project. Here's a video on how neural-nets work that uses the character recognition.

Thanks. I am still novice in the field, and it seems a little complicated to do these things now. Neural networks/deep learning is a topic on its own. This leads me to the following question: What are employers looking for in the personal projects? What do I need to demonstrate as skills? How complex my project should be? Thanks in advance

 

[donpacino]

The ultimate goal of a field like data science is to draw conclusion from data. Ideally you will be able to make some conclusion or recommendation from the data that you have.

As an example. Let's say the data-set is related to insurance, age, current estimated risk, number of accidents, cost, type of vehicle, etc...

An employer would look that you can take a block of data and draw a conclusion from it. It is not enough to say here is the difference in number of accidents between age 85 and 75 drivers. You need to say "The price of insurance for 75 year old needs to go up due to the fact that the profit after paying out for accidents is not high enough." You need to develop a program/method that will essentially fill in that data for you. You need to determine a way to evaluate through coding if an age group or car group is worth it. A good way to start would be just looking at histograms and scatter plots, then start thinking of ways to automatically identify those trends.

TLDR: with your dataset, develop ways to identify trends, then think about what those trends mean, and report on it.