I've recently finished my computer science bachelors, I want to get into data-science / machine learning and am looking for a project to do in my spare time.
I'm hoping to create a decision support tool which could possibly make me some money if possible, using data-mining algorithms (Predicting specific social changes? )
I know that football games can be predicted with the rough estimation of 70% accuracy, i'm looking for something a little more certain than that. My final year project, I used multiple databases of patients and generated a classification model to predict the chance of a new patient developing colon-cancer.
Anyone worked with big-data / data-analytics or data-mining who can shed some light or give me some useful project ideas?
I'm looking to do everything in python / C# but could probably pick up R pretty easily.
Any-one have any thoughts?
Thanks in advance
>>61245640
Stick with python, C# and R are garbage
numpy, scipy and panda along with shit tons of deep learning libraries are all you need
>>61245666
Yeah, I've been doing some bits with numpy and padas it's what i'd most likely be using,
I mainly put C# in there for a front-end sorta thing and due to being slightly more comfortable using it than any other languages.
>>61245733
With the exception of GUI and multithreading/multiprocessing anything you can do in C# you can whip up something equivalent in Python much faster. With a little bit of luck it might even run faster if you use the native libraries right.
Jupyter Notebook is also a really good IDE substitute that allow you to prototype and iterate very fast, which is crucial in data science
I was a C# programmer before I code almost exclusively in Python now.
>>61245835
Thank you for that recommendation. I had never heard of Jupyter before it's pretty handy
>>61246004
Here's a list of my cookie cutter python libraries for data science:
numpy
scipy
pandas
scikit-learn
xgboost
nltk
[deep learning library of your choice]
matplotlib + seaborn
pickle
tqdm
You can try your hands on some Kaggle competitions if you've got the time
>>61246102
Much appreciated!