Building a data scientist portfolio, that stands out from crowd, is very important if you are looking to get a job in this field. Internet is full of data science projects. You cannot make your portfolio look authentic and different by doing what is already flooding on internet. You must do things that are different and have potential to bring some value to society.
In this article, I'll share 7 data science project ideas, that can make your portfolio different and add weightage.
1. Handwritten equation solver using CNN
Recognizing handwritings are one of the tasks that can bring a lot of value to everyone, It can be used in banks hospitals, railway counters etc, to recognize data filled by customers on paper forms. Computer vision and ML is used to recognize handwritings.
Recognizing mathematical expressions is very confusing and rigorous task in ML, having this on your portfolio will bring a lot of weight.
2. Business meeting summary generation using NLP
To summarize any given text you need to understand context from it. This has been a great challenge for data scientists in NLP. To implement a meeting summary generator you can use NLTK of python and other scipy packages. This project will showcase your authentic work in NLP.
It also does have a great potential.
3. Facial recognition to detect mood and play song accordingly.
Detecting mood from a face is a challenge, if you implement it you will be challenged to do a lot of work on your own. Second thing is, you have to play musics also, from a certain flavour, so, it would require you to do taggings to musics also.
After tagging musics, based upon their lyrics, you now need to pipeline everything together to make the whole thing work as a mood maker.
4. Detecting depression using social media post sentiment analysis
Doing a sentiment analysis will require you to implement some concepts from NLP, and detecting depression out of that information will require you to build another model. Also, connecting with social media and fetching data from it will force you to learn to use a lot social media apis, data cleaning and other important stuffs. This is a great project to have on your portfolio if you want to showcase your NLP and data science skills.
5. Image regeneration for old damaged reel picture
Constructing good looking images from old damaged pictures is a challenge. It will require you to do a lot of play arounds with GNN and other data science libraries. If you implement this project, it will be a great tool to have for your personal use and to showcase your data science skills also. From user's perspective, people do visit shops and pay to get good images from any damaged reel, having that thing simplified will be very valuable.
6. Music generation using deep learning
Musics are a set of amplitudes and frequencies, when put together at a certain tempo, they sound good to our brain. So, Basically, they are a series of numbers. You can work around deep learning, build a model, train it with a lot of musics and make it come up with its own music.
Though, this is very challenging task to implement, but having that on your portfolio is certainly a great thing.
7. Doing image analysis on images captured from space telescopes
This is a very adventures thing to do. Space itself if full of mysteries. Several space agencies across world publish images captured from their space vehicles. You can do analysis on those images to detect extraterrestrial lives, new stars, planets and so many things. In order to do that, however, you will have to get some knowledge about what properties depict what in the image.
Doing analysis on space images, is a very practical thing. It adds a lot of power to your portfolio.