Mini project 1 done during my first semester at SFSU in data mining class
- python3 classifier.py (default mode)
- --discretize, --no-discretize: a. discretize numeric values using equal bins (default: True)
- --use_mode, --no-use_mode a. use mode or mean to fill NaN data (default: True)
python3 classifier.pyThe adult.csv dataset is used
I used Weka software to analyse the dataset.
At the beginning, i started to work on data analysis using matplotlib on jupyter notebook but it wasn’t efficient so i used weka.
To start this project I had to understand what is naive bayes. So I started my research and started an algorithm without success. I discovered that to make machine learning I had to work on the data before doing anything.
I created a function to clean the data and split the dataset. After that I tried to restart the naive bayes algo and get a result. Then I added next features like flags and what was wrong.
Finally I created the k fold cross validation function.
My results are not too bad for this first project. I obtained 0.73 for the best accuracy score. I can upgrade the algorithm to better fit and train the model. I can also upgrade the rapidity of execution.